1
-*- coding: utf-8; fill-column: 68 -*-
10
- Fixed erroneous self_group() call in expression package.
16
- The copy() method on Column now supports uninitialized,
17
unnamed Column objects. This allows easy creation of
18
declarative helpers which place common columns on multiple
21
- Default generators like Sequence() translate correctly
22
across a copy() operation.
24
- Sequence() and other DefaultGenerator objects are accepted
25
as the value for the "default" and "onupdate" keyword
26
arguments of Column, in addition to being accepted
29
- Fixed a column arithmetic bug that affected column
30
correspondence for cloned selectables which contain
31
free-standing column expressions. This bug is
32
generally only noticeable when exercising newer
33
ORM behavior only availble in 0.6 via [ticket:1568],
34
but is more correct at the SQL expression level
35
as well. [ticket:1617]
38
- The extract() function, which was slightly improved in
39
0.5.7, needed a lot more work to generate the correct
40
typecast (the typecasts appear to be necessary in PG's
41
EXTRACT quite a lot of the time). The typecast is
42
now generated using a rule dictionary based
43
on PG's documentation for date/time/interval arithmetic.
44
It also accepts text() constructs again, which was broken
45
in 0.5.7. [ticket:1647]
48
- Recognize more errors as disconnections. [ticket:1646]
53
- contains_eager() now works with the automatically
54
generated subquery that results when you say
55
"query(Parent).join(Parent.somejoinedsubclass)", i.e.
56
when Parent joins to a joined-table-inheritance subclass.
57
Previously contains_eager() would erroneously add the
58
subclass table to the query separately producing a
59
cartesian product. An example is in the ticket
60
description. [ticket:1543]
62
- query.options() now only propagate to loaded objects
63
for potential further sub-loads only for options where
64
such behavior is relevant, keeping
65
various unserializable options like those generated
66
by contains_eager() out of individual instance states.
69
- Session.execute() now locates table- and
70
mapper-specific binds based on a passed
71
in expression which is an insert()/update()/delete()
72
construct. [ticket:1054]
74
- Session.merge() now properly overwrites a many-to-one or
75
uselist=False attribute to None if the attribute
76
is also None in the given object to be merged.
78
- Fixed a needless select which would occur when merging
79
transient objects that contained a null primary key
80
identifier. [ticket:1618]
82
- Mutable collection passed to the "extension" attribute
83
of relation(), column_property() etc. will not be mutated
84
or shared among multiple instrumentation calls, preventing
85
duplicate extensions, such as backref populators,
86
from being inserted into the list.
89
- Fixed the call to get_committed_value() on CompositeProperty.
92
- Fixed bug where Query would crash if a join() with no clear
93
"left" side were called when a non-mapped column entity
94
appeared in the columns list. [ticket:1602]
96
- Fixed bug whereby composite columns wouldn't load properly
97
when configured on a joined-table subclass, introduced in
98
version 0.5.6 as a result of the fix for [ticket:1480].
99
[ticket:1616] thx to Scott Torborg.
101
- The "use get" behavior of many-to-one relations, i.e. that a
102
lazy load will fallback to the possibly cached query.get()
103
value, now works across join conditions where the two compared
104
types are not exactly the same class, but share the same
105
"affinity" - i.e. Integer and SmallInteger. Also allows
106
combinations of reflected and non-reflected types to work
107
with 0.5 style type reflection, such as PGText/Text (note 0.6
108
reflects types as their generic versions). [ticket:1556]
110
- Fixed bug in query.update() when passing Cls.attribute
111
as keys in the value dict and using synchronize_session='expire'
112
('fetch' in 0.6). [ticket:1436]
115
- Fixed bug in two-phase transaction whereby commit() method
116
didn't set the full state which allows subsequent close()
117
call to succeed. [ticket:1603]
119
- Fixed the "numeric" paramstyle, which apparently is the
120
default paramstyle used by Informixdb.
122
- Repeat expressions in the columns clause of a select
123
are deduped based on the identity of each clause element,
124
not the actual string. This allows positional
125
elements to render correctly even if they all render
126
identically, such as "qmark" style bind parameters.
129
- The cursor associated with connection pool connections
130
(i.e. _CursorFairy) now proxies `__iter__()` to the
131
underlying cursor correctly. [ticket:1632]
133
- types now support an "affinity comparison" operation, i.e.
134
that an Integer/SmallInteger are "compatible", or
135
a Text/String, PickleType/Binary, etc. Part of
138
- Fixed bug preventing alias() of an alias() from being
139
cloned or adapted (occurs frequently in ORM operations).
143
- sqlite dialect properly generates CREATE INDEX for a table
144
that is in an alternate schema. [ticket:1439]
147
- Added support for reflecting the DOUBLE PRECISION type,
148
via a new postgres.PGDoublePrecision object.
149
This is postgresql.DOUBLE_PRECISION in 0.6.
152
- Added support for reflecting the INTERVAL YEAR TO MONTH
153
and INTERVAL DAY TO SECOND syntaxes of the INTERVAL
156
- Corrected the "has_sequence" query to take current schema,
157
or explicit sequence-stated schema, into account.
160
- Fixed the behavior of extract() to apply operator
161
precedence rules to the "::" operator when applying
162
the "timestamp" cast - ensures proper parenthesization.
166
- Changed the name of TrustedConnection to
167
Trusted_Connection when constructing pyodbc connect
168
arguments [ticket:1561]
171
- The "table_names" dialect function, used by MetaData
172
.reflect(), omits "index overflow tables", a system
173
table generated by Oracle when "index only tables"
174
with overflow are used. These tables aren't accessible
175
via SQL and can't be reflected. [ticket:1637]
178
- A column can be added to a joined-table declarative
179
superclass after the class has been constructed
180
(i.e. via class-level attribute assignment), and
181
the column will be propagated down to
182
subclasses. [ticket:1570] This is the reverse
183
situation as that of [ticket:1523], fixed in 0.5.6.
185
- Fixed a slight inaccuracy in the sharding example.
186
Comparing equivalence of columns in the ORM is best
187
accomplished using col1.shares_lineage(col2).
190
- Removed unused `load()` method from ShardedQuery.
196
- Fixed bug whereby inheritance discriminator part of a
197
composite primary key would fail on updates.
198
Continuation of [ticket:1300].
200
- Fixed bug which disallowed one side of a many-to-many
201
bidirectional reference to declare itself as "viewonly"
204
- Added an assertion that prevents a @validates function
205
or other AttributeExtension from loading an unloaded
206
collection such that internal state may be corrupted.
209
- Fixed bug which prevented two entities from mutually
210
replacing each other's primary key values within a single
211
flush() for some orderings of operations. [ticket:1519]
213
- Fixed an obscure issue whereby a joined-table subclass
214
with a self-referential eager load on the base class
215
would populate the related object's "subclass" table with
216
data from the "subclass" table of the parent.
219
- relations() now have greater ability to be "overridden",
220
meaning a subclass that explicitly specifies a relation()
221
overriding that of the parent class will be honored
222
during a flush. This is currently to support
223
many-to-many relations from concrete inheritance setups.
224
Outside of that use case, YMMV. [ticket:1477]
226
- Squeezed a few more unnecessary "lazy loads" out of
227
relation(). When a collection is mutated, many-to-one
228
backrefs on the other side will not fire off to load
229
the "old" value, unless "single_parent=True" is set.
230
A direct assignment of a many-to-one still loads
231
the "old" value in order to update backref collections
232
on that value, which may be present in the session
233
already, thus maintaining the 0.5 behavioral contract.
236
- Fixed bug whereby a load/refresh of joined table
237
inheritance attributes which were based on
238
column_property() or similar would fail to evaluate.
241
- Improved support for MapperProperty objects overriding
242
that of an inherited mapper for non-concrete
243
inheritance setups - attribute extensions won't randomly
244
collide with each other. [ticket:1488]
246
- UPDATE and DELETE do not support ORDER BY, LIMIT, OFFSET,
247
etc. in standard SQL. Query.update() and Query.delete()
248
now raise an exception if any of limit(), offset(),
249
order_by(), group_by(), or distinct() have been
250
called. [ticket:1487]
252
- Added AttributeExtension to sqlalchemy.orm.__all__
254
- Improved error message when query() is called with
255
a non-SQL /entity expression. [ticket:1476]
257
- Using False or 0 as a polymorphic discriminator now
258
works on the base class as well as a subclass.
261
- Added enable_assertions(False) to Query which disables
262
the usual assertions for expected state - used
263
by Query subclasses to engineer custom state.
265
http://www.sqlalchemy.org/trac/wiki/UsageRecipes/PreFilteredQuery
268
- Fixed recursion issue which occured if a mapped object's
269
`__len__()` or `__nonzero__()` method resulted in state
270
changes. [ticket:1501]
272
- Fixed incorrect exception raise in
273
Weak/StrongIdentityMap.add()
276
- Fixed the error message for "could not find a FROM clause"
277
in query.join() which would fail to issue correctly
278
if the query was against a pure SQL construct.
281
- Fixed a somewhat hypothetical issue which would result
282
in the wrong primary key being calculated for a mapper
283
using the old polymorphic_union function - but this
284
is old stuff. [ticket:1486]
287
- Fixed column.copy() to copy defaults and onupdates.
290
- Fixed a bug in extract() introduced in 0.5.4 whereby
291
the string "field" argument was getting treated as a
292
ClauseElement, causing various errors within more
293
complex SQL transformations.
295
- Unary expressions such as DISTINCT propagate their
296
type handling to result sets, allowing conversions like
297
unicode and such to take place. [ticket:1420]
299
- Fixed bug in Table and Column whereby passing empty
300
dict for "info" argument would raise an exception.
304
- Backported 0.6 fix for Oracle alias names not getting
305
truncated. [ticket:1309]
308
- The collection proxies produced by associationproxy are now
309
pickleable. A user-defined proxy_factory however
310
is still not pickleable unless it defines __getstate__
311
and __setstate__. [ticket:1446]
313
- Declarative will raise an informative exception if
314
__table_args__ is passed as a tuple with no dict argument.
315
Improved documentation. [ticket:1468]
317
- Table objects declared in the MetaData can now be used
318
in string expressions sent to primaryjoin/secondaryjoin/
319
secondary - the name is pulled from the MetaData of the
320
declarative base. [ticket:1527]
322
- A column can be added to a joined-table subclass after
323
the class has been constructed (i.e. via class-level
324
attribute assignment). The column is added to the underlying
325
Table as always, but now the mapper will rebuild its
326
"join" to include the new column, instead of raising
327
an error about "no such column, use column_property()
328
instead". [ticket:1523]
331
- Added examples into the test suite so they get exercised
332
regularly and cleaned up a couple deprecation warnings.
338
- unit tests have been migrated from unittest to nose. See
339
README.unittests for information on how to run the tests.
343
- The "foreign_keys" argument of relation() will now propagate
344
automatically to the backref in the same way that primaryjoin
345
and secondaryjoin do. For the extremely rare use case where
346
the backref of a relation() has intentionally different
347
"foreign_keys" configured, both sides now need to be
348
configured explicity (if they do in fact require this setting,
349
see the next note...).
351
- ...the only known (and really, really rare) use case where a
352
different foreign_keys setting was used on the
353
forwards/backwards side, a composite foreign key that
354
partially points to its own columns, has been enhanced such
355
that the fk->itself aspect of the relation won't be used to
356
determine relation direction.
358
- Session.mapper is now *deprecated*.
360
Call session.add() if you'd like a free-standing object to be
361
part of your session. Otherwise, a DIY version of
362
Session.mapper is now documented at
363
http://www.sqlalchemy.org/trac/wiki/UsageRecipes/SessionAwareMapper
364
The method will remain deprecated throughout 0.6.
366
- Fixed Query being able to join() from individual columns of a
367
joined-table subclass entity, i.e. query(SubClass.foo,
368
SubcClass.bar).join(<anything>). In most cases, an error
369
"Could not find a FROM clause to join from" would be
370
raised. In a few others, the result would be returned in terms
371
of the base class rather than the subclass - so applications
372
which relied on this erroneous result need to be
373
adjusted. [ticket:1431]
375
- Fixed a bug involving contains_eager(), which would apply
376
itself to a secondary (i.e. lazy) load in a particular rare
377
case, producing cartesian products. improved the targeting of
378
query.options() on secondary loads overall [ticket:1461].
380
- Fixed bug introduced in 0.5.4 whereby Composite types fail
381
when default-holding columns are flushed.
383
- Fixed another 0.5.4 bug whereby mutable attributes
384
(i.e. PickleType) wouldn't be deserialized correctly when the
385
whole object was serialized. [ticket:1426]
387
- Fixed bug whereby session.is_modified() would raise an
388
exception if any synonyms were in use.
390
- Fixed potential memory leak whereby previously pickled objects
391
placed back in a session would not be fully garbage collected
392
unless the Session were explicitly closed out.
394
- Fixed bug whereby list-based attributes, like pickletype and
395
PGArray, failed to be merged() properly.
397
- Repaired non-working attributes.set_committed_value function.
399
- Trimmed the pickle format for InstanceState which should
400
further reduce the memory footprint of pickled instances. The
401
format should be backwards compatible with that of 0.5.4 and
404
- sqlalchemy.orm.join and sqlalchemy.orm.outerjoin are now
405
added to __all__ in sqlalchemy.orm.*. [ticket:1463]
407
- Fixed bug where Query exception raise would fail when
408
a too-short composite primary key value were passed to
412
- Removed an obscure feature of execute() (including connection,
413
engine, Session) whereby a bindparam() construct can be sent
414
as a key to the params dictionary. This usage is undocumented
415
and is at the core of an issue whereby the bindparam() object
416
created implicitly by a text() construct may have the same
417
hash value as a string placed in the params dictionary and may
418
result in an inappropriate match when computing the final bind
419
parameters. Internal checks for this condition would add
420
significant latency to the critical task of parameter
421
rendering, so the behavior is removed. This is a backwards
422
incompatible change for any application that may have been
423
using this feature, however the feature has never been
427
- Implemented recreate() for StaticPool.
434
- Repaired the printing of SQL exceptions which are not
435
based on parameters or are not executemany() style.
438
- Deprecated the hardcoded TIMESTAMP function, which when
439
used as func.TIMESTAMP(value) would render "TIMESTAMP value".
440
This breaks on some platforms as PostgreSQL doesn't allow
441
bind parameters to be used in this context. The hard-coded
442
uppercase is also inappropriate and there's lots of other
443
PG casts that we'd need to support. So instead, use
444
text constructs i.e. select(["timestamp '12/05/09'"]).
451
- Fixed an attribute error introduced in 0.5.4 which would
452
occur when merge() was used with an incomplete object.
458
- Significant performance enhancements regarding Sessions/flush()
459
in conjunction with large mapper graphs, large numbers of
462
- Removed all* O(N) scanning behavior from the flush() process,
463
i.e. operations that were scanning the full session,
464
including an extremely expensive one that was erroneously
465
assuming primary key values were changing when this
468
* one edge case remains which may invoke a full scan,
469
if an existing primary key attribute is modified
472
- The Session's "weak referencing" behavior is now *full* -
473
no strong references whatsoever are made to a mapped object
474
or related items/collections in its __dict__. Backrefs and
475
other cycles in objects no longer affect the Session's ability
476
to lose all references to unmodified objects. Objects with
477
pending changes still are maintained strongly until flush.
480
The implementation also improves performance by moving
481
the "resurrection" process of garbage collected items
482
to only be relevant for mappings that map "mutable"
483
attributes (i.e. PickleType, composite attrs). This removes
484
overhead from the gc process and simplifies internal
487
If a "mutable" attribute change is the sole change on an object
488
which is then dereferenced, the mapper will not have access to
489
other attribute state when the UPDATE is issued. This may present
490
itself differently to some MapperExtensions.
492
The change also affects the internal attribute API, but not
493
the AttributeExtension interface nor any of the publically
494
documented attribute functions.
496
- The unit of work no longer genererates a graph of "dependency"
497
processors for the full graph of mappers during flush(), instead
498
creating such processors only for those mappers which represent
499
objects with pending changes. This saves a tremendous number
500
of method calls in the context of a large interconnected
503
- Cached a wasteful "table sort" operation that previously
504
occured multiple times per flush, also removing significant
505
method call count from flush().
507
- Other redundant behaviors have been simplified in
510
- Modified query_cls on DynamicAttributeImpl to accept a full
511
mixin version of the AppenderQuery, which allows subclassing
514
- The "polymorphic discriminator" column may be part of a
515
primary key, and it will be populated with the correct
516
discriminator value. [ticket:1300]
518
- Fixed the evaluator not being able to evaluate IS NULL clauses.
520
- Fixed the "set collection" function on "dynamic" relations to
521
initiate events correctly. Previously a collection could only
522
be assigned to a pending parent instance, otherwise modified
523
events would not be fired correctly. Set collection is now
524
compatible with merge(), fixes [ticket:1352].
526
- Allowed pickling of PropertyOption objects constructed with
527
instrumented descriptors; previously, pickle errors would occur
528
when pickling an object which was loaded with a descriptor-based
529
option, such as query.options(eagerload(MyClass.foo)).
531
- Lazy loader will not use get() if the "lazy load" SQL clause
532
matches the clause used by get(), but contains some parameters
533
hardcoded. Previously the lazy strategy would fail with the
534
get(). Ideally get() would be used with the hardcoded
535
parameters but this would require further development.
538
- MapperOptions and other state associated with query.options()
539
is no longer bundled within callables associated with each
540
lazy/deferred-loading attribute during a load.
541
The options are now associated with the instance's
542
state object just once when it's populated. This removes
543
the need in most cases for per-instance/attribute loader
544
objects, improving load speed and memory overhead for
545
individual instances. [ticket:1391]
547
- Fixed another location where autoflush was interfering
548
with session.merge(). autoflush is disabled completely
549
for the duration of merge() now. [ticket:1360]
551
- Fixed bug which prevented "mutable primary key" dependency
552
logic from functioning properly on a one-to-one
553
relation(). [ticket:1406]
555
- Fixed bug in relation(), introduced in 0.5.3,
556
whereby a self referential relation
557
from a base class to a joined-table subclass would
558
not configure correctly.
560
- Fixed obscure mapper compilation issue when inheriting
561
mappers are used which would result in un-initialized
564
- Fixed documentation for session weak_identity_map -
565
the default value is True, indicating a weak
566
referencing map in use.
568
- Fixed a unit of work issue whereby the foreign
569
key attribute on an item contained within a collection
570
owned by an object being deleted would not be set to
571
None if the relation() was self-referential. [ticket:1376]
573
- Fixed Query.update() and Query.delete() failures with eagerloaded
574
relations. [ticket:1378]
576
- It is now an error to specify both columns of a binary primaryjoin
577
condition in the foreign_keys or remote_side collection. Whereas
578
previously it was just nonsensical, but would succeed in a
579
non-deterministic way.
582
- Added a quote_schema() method to the IdentifierPreparer class
583
so that dialects can override how schemas get handled. This
584
enables the MSSQL dialect to treat schemas as multipart
585
identifiers, such as 'database.owner'. [ticket: 594, 1341]
588
- Back-ported the "compiler" extension from SQLA 0.6. This
589
is a standardized interface which allows the creation of custom
590
ClauseElement subclasses and compilers. In particular it's
591
handy as an alternative to text() when you'd like to
592
build a construct that has database-specific compilations.
593
See the extension docs for details.
595
- Exception messages are truncated when the list of bound
596
parameters is larger than 10, preventing enormous
597
multi-page exceptions from filling up screens and logfiles
598
for large executemany() statements. [ticket:1413]
600
- ``sqlalchemy.extract()`` is now dialect sensitive and can
601
extract components of timestamps idiomatically across the
602
supported databases, including SQLite.
604
- Fixed __repr__() and other _get_colspec() methods on
605
ForeignKey constructed from __clause_element__() style
606
construct (i.e. declarative columns). [ticket:1353]
609
- Reflecting a FOREIGN KEY construct will take into account
610
a dotted schema.tablename combination, if the foreign key
611
references a table in a remote schema. [ticket:1405]
614
- Modified how savepoint logic works to prevent it from
615
stepping on non-savepoint oriented routines. Savepoint
616
support is still very experimental.
618
- Added in reserved words for MSSQL that covers version 2008
619
and all prior versions. [ticket:1310]
621
- Corrected problem with information schema not working with a
622
binary collation based database. Cleaned up information schema
623
since it is only used by mssql now. [ticket:1343]
626
- Corrected the SLBoolean type so that it properly treats only 1
627
as True. [ticket:1402]
629
- Corrected the float type so that it correctly maps to a
630
SLFloat type when being reflected. [ticket:1273]
634
- Fixed adding of deferred or other column properties to a
635
declarative class. [ticket:1379]
640
- The "objects" argument to session.flush() is deprecated.
641
State which represents the linkage between a parent and
642
child object does not support "flushed" status on
643
one side of the link and not the other, so supporting
644
this operation leads to misleading results.
647
- Query now implements __clause_element__() which produces
648
its selectable, which means a Query instance can be accepted
649
in many SQL expressions, including col.in_(query),
650
union(query1, query2), select([foo]).select_from(query),
653
- Query.join() can now construct multiple FROM clauses, if
654
needed. Such as, query(A, B).join(A.x).join(B.y)
655
might say SELECT A.*, B.* FROM A JOIN X, B JOIN Y.
656
Eager loading can also tack its joins onto those
657
multiple FROM clauses. [ticket:1337]
659
- Fixed bug in dynamic_loader() where append/remove events
660
after construction time were not being propagated to the
661
UOW to pick up on flush(). [ticket:1347]
663
- Fixed bug where column_prefix wasn't being checked before
664
not mapping an attribute that already had class-level
667
- a session.expire() on a particular collection attribute
668
will clear any pending backref additions as well, so that
669
the next access correctly returns only what was present
670
in the database. Presents some degree of a workaround for
671
[ticket:1315], although we are considering removing the
672
flush([objects]) feature altogether.
674
- Session.scalar() now converts raw SQL strings to text()
675
the same way Session.execute() does and accepts same
676
alternative **kw args.
678
- improvements to the "determine direction" logic of
679
relation() such that the direction of tricky situations
680
like mapper(A.join(B)) -> relation-> mapper(B) can be
683
- When flushing partial sets of objects using session.flush([somelist]),
684
pending objects which remain pending after the operation won't
685
inadvertently be added as persistent. [ticket:1306]
687
- Added "post_configure_attribute" method to InstrumentationManager,
688
so that the "listen_for_events.py" example works again.
691
- a forward and complementing backwards reference which are both
692
of the same direction, i.e. ONETOMANY or MANYTOONE,
693
is now detected, and an error message is raised.
694
Saves crazy CircularDependencyErrors later on.
696
- Fixed bugs in Query regarding simultaneous selection of
697
multiple joined-table inheritance entities with common base
700
- previously the adaption applied to "B" on
701
"A JOIN B" would be erroneously partially applied
704
- comparisons on relations (i.e. A.related==someb)
705
were not getting adapted when they should.
707
- Other filterings, like
708
query(A).join(A.bs).filter(B.foo=='bar'), were erroneously
709
adapting "B.foo" as though it were an "A".
711
- Fixed adaptation of EXISTS clauses via any(), has(), etc.
712
in conjunction with an aliased object on the left and
713
of_type() on the right. [ticket:1325]
715
- Added an attribute helper method ``set_committed_value`` in
716
sqlalchemy.orm.attributes. Given an object, attribute name,
717
and value, will set the value on the object as part of its
718
"committed" state, i.e. state that is understood to have
719
been loaded from the database. Helps with the creation of
720
homegrown collection loaders and such.
722
- Query won't fail with weakref error when a non-mapper/class
723
instrumented descriptor is passed, raises
724
"Invalid column expession".
726
- Query.group_by() properly takes into account aliasing applied
727
to the FROM clause, such as with select_from(), using
728
with_polymorphic(), or using from_self().
731
- An alias() of a select() will convert to a "scalar subquery"
732
when used in an unambiguously scalar context, i.e. it's used
733
in a comparison operation. This applies to
734
the ORM when using query.subquery() as well.
736
- Fixed missing _label attribute on Function object, others
737
when used in a select() with use_labels (such as when used
738
in an ORM column_property()). [ticket:1302]
740
- anonymous alias names now truncate down to the max length
741
allowed by the dialect. More significant on DBs like
742
Oracle with very small character limits. [ticket:1309]
744
- the __selectable__() interface has been replaced entirely
745
by __clause_element__().
747
- The per-dialect cache used by TypeEngine to cache
748
dialect-specific types is now a WeakKeyDictionary.
749
This to prevent dialect objects from
750
being referenced forever for an application that
751
creates an arbitrarily large number of engines
752
or dialects. There is a small performance penalty
753
which will be resolved in 0.6. [ticket:1299]
756
- Fixed SQLite reflection methods so that non-present
757
cursor.description, which triggers an auto-cursor
758
close, will be detected so that no results doesn't
759
fail on recent versions of pysqlite which raise
760
an error when fetchone() called with no rows present.
763
- Index reflection won't fail when an index with
764
multiple expressions is encountered.
766
- Added PGUuid and PGBit types to
767
sqlalchemy.databases.postgres. [ticket:1327]
769
- Refection of unknown PG types won't crash when those
770
types are specified within a domain. [ticket:1327]
773
- Preliminary support for pymssql 1.0.1
775
- Corrected issue on mssql where max_identifier_length was
780
- Fixed a recursive pickling issue in serializer, triggered
781
by an EXISTS or other embedded FROM construct.
783
- Declarative locates the "inherits" class using a search
784
through __bases__, to skip over mixins that are local
787
- Declarative figures out joined-table inheritance primary join
788
condition even if "inherits" mapper argument is given
791
- Declarative will properly interpret the "foreign_keys" argument
792
on a backref() if it's a string.
794
- Declarative will accept a table-bound column as a property
795
when used in conjunction with __table__, if the column is already
796
present in __table__. The column will be remapped to the given
797
key the same way as when added to the mapper() properties dict.
803
- Further refined 0.5.1's warning about delete-orphan cascade
804
placed on a many-to-many relation. First, the bad news:
805
the warning will apply to both many-to-many as well as
806
many-to-one relations. This is necessary since in both
807
cases, SQLA does not scan the full set of potential parents
808
when determining "orphan" status - for a persistent object
809
it only detects an in-python de-association event to establish
810
the object as an "orphan". Next, the good news: to support
811
one-to-one via a foreign key or assocation table, or to
812
support one-to-many via an association table, a new flag
813
single_parent=True may be set which indicates objects
814
linked to the relation are only meant to have a single parent.
815
The relation will raise an error if multiple parent-association
816
events occur within Python.
818
- Adjusted the attribute instrumentation change from 0.5.1 to
819
fully establish instrumentation for subclasses where the mapper
820
was created after the superclass had already been fully
821
instrumented. [ticket:1292]
823
- Fixed bug in delete-orphan cascade whereby two one-to-one
824
relations from two different parent classes to the same target
825
class would prematurely expunge the instance.
827
- Fixed an eager loading bug whereby self-referential eager
828
loading would prevent other eager loads, self referential or not,
829
from joining to the parent JOIN properly. Thanks to Alex K
830
for creating a great test case.
832
- session.expire() and related methods will not expire() unloaded
833
deferred attributes. This prevents them from being needlessly
834
loaded when the instance is refreshed.
836
- query.join()/outerjoin() will now properly join an aliased()
837
construct to the existing left side, even if query.from_self()
838
or query.select_from(someselectable) has been called.
842
- Further fixes to the "percent signs and spaces in column/table
843
names" functionality. [ticket:1284]
846
- Restored convert_unicode handling. Results were being passed
847
on through without conversion. [ticket:1291]
849
- Really fixing the decimal handling this time. [ticket:1282].
851
- Modified table reflection code to use only kwargs when
852
constructing tables. [Ticket:1289]
858
- Removed an internal join cache which could potentially leak
859
memory when issuing query.join() repeatedly to ad-hoc
862
- The "clear()", "save()", "update()", "save_or_update()"
863
Session methods have been deprecated, replaced by
864
"expunge_all()" and "add()". "expunge_all()" has also
865
been added to ScopedSession.
867
- Modernized the "no mapped table" exception and added a more
868
explicit __table__/__tablename__ exception to declarative.
870
- Concrete inheriting mappers now instrument attributes which
871
are inherited from the superclass, but are not defined for
872
the concrete mapper itself, with an InstrumentedAttribute that
873
issues a descriptive error when accessed. [ticket:1237]
875
- Added a new `relation()` keyword `back_populates`. This
876
allows configuation of backreferences using explicit
877
relations. [ticket:781] This is required when creating
878
bidirectional relations between a hierarchy of concrete
879
mappers and another class. [ticket:1237]
881
- Test coverage added for `relation()` objects specified on
882
concrete mappers. [ticket:1237]
884
- Query.from_self() as well as query.subquery() both disable
885
the rendering of eager joins inside the subquery produced.
886
The "disable all eager joins" feature is available publically
887
via a new query.enable_eagerloads() generative. [ticket:1276]
889
- Added a rudimental series of set operations to Query that
890
receive Query objects as arguments, including union(),
891
union_all(), intersect(), except_(), insertsect_all(),
892
except_all(). See the API documentation for
893
Query.union() for examples.
895
- Fixed bug that prevented Query.join() and eagerloads from
896
attaching to a query that selected from a union or aliased union.
898
- A short documentation example added for bidirectional
899
relations specified on concrete mappers. [ticket:1237]
901
- Mappers now instrument class attributes upon construction
902
with the final InstrumentedAttribute object which remains
903
persistent. The `_CompileOnAttr`/`__getattribute__()`
904
methodology has been removed. The net effect is that
905
Column-based mapped class attributes can now be used fully
906
at the class level without invoking a mapper compilation
907
operation, greatly simplifying typical usage patterns
908
within declarative. [ticket:1269]
910
- ColumnProperty (and front-end helpers such as ``deferred``) no
911
longer ignores unknown **keyword arguments.
913
- Fixed a bug with the unitofwork's "row switch" mechanism,
914
i.e. the conversion of INSERT/DELETE into an UPDATE, when
915
combined with joined-table inheritance and an object
916
which contained no defined values for the child table where
917
an UPDATE with no SET clause would be rendered.
919
- Using delete-orphan on a many-to-many relation is deprecated.
920
This produces misleading or erroneous results since SQLA does
921
not retrieve the full list of "parents" for m2m. To get delete-orphan
922
behavior with an m2m table, use an explcit association class
923
so that the individual association row is treated as a parent.
926
- delete-orphan cascade always requires delete cascade. Specifying
927
delete-orphan without delete now raises a deprecation warning.
931
- Improved the methodology to handling percent signs in column
932
names from [ticket:1256]. Added more tests. MySQL and
933
PostgreSQL dialects still do not issue correct CREATE TABLE
934
statements for identifiers with percent signs in them.
937
- Index now accepts column-oriented InstrumentedAttributes
938
(i.e. column-based mapped class attributes) as column
939
arguments. [ticket:1214]
941
- Column with no name (as in declarative) won't raise a
942
NoneType error when it's string output is requsted
943
(such as in a stack trace).
945
- Fixed bug when overriding a Column with a ForeignKey
946
on a reflected table, where derived columns (i.e. the
947
"virtual" columns of a select, etc.) would inadvertently
948
call upon schema-level cleanup logic intended only
949
for the original column. [ticket:1278]
952
- Can now specify Column objects on subclasses which have no
953
table of their own (i.e. use single table inheritance).
954
The columns will be appended to the base table, but only
955
mapped by the subclass.
957
- For both joined and single inheriting subclasses, the subclass
958
will only map those columns which are already mapped on the
959
superclass and those explicit on the subclass. Other
960
columns that are present on the `Table` will be excluded
961
from the mapping by default, which can be disabled
962
by passing a blank `exclude_properties` collection to the
963
`__mapper_args__`. This is so that single-inheriting
964
classes which define their own columns are the only classes
965
to map those columns. The effect is actually a more organized
966
mapping than you'd normally get with explicit `mapper()`
967
calls unless you set up the `exclude_properties` arguments
970
- It's an error to add new Column objects to a declarative class
971
that specified an existing table using __table__.
974
- Added the missing keywords from MySQL 4.1 so they get escaped
978
- Corrected handling of large decimal values with more robust
979
tests. Removed string manipulation on floats. [ticket:1280]
981
- Modified the do_begin handling in mssql to use the Cursor not
982
the Connection so it is DBAPI compatible.
984
- Corrected SAVEPOINT support on adodbapi by changing the
985
handling of savepoint_release, which is unsupported on mssql.
992
- Documentation has been converted to Sphinx. In particular,
993
the generated API documentation has been constructed into a
994
full blown "API Reference" section which organizes editorial
995
documentation combined with generated docstrings. Cross
996
linking between sections and API docs are vastly improved, a
997
javascript-powered search feature is provided, and a full
998
index of all classes, functions and members is provided.
1000
- setup.py now imports setuptools only optionally. If not
1001
present, distutils is used. The new "pip" installer is
1002
recommended over easy_install as it installs in a more
1005
- added an extremely basic illustration of a PostGIS integration
1006
to the examples folder.
1009
- Query.with_polymorphic() now accepts a third argument
1010
"discriminator" which will replace the value of
1011
mapper.polymorphic_on for that query. Mappers themselves no
1012
longer require polymorphic_on to be set, even if the mapper
1013
has a polymorphic_identity. When not set, the mapper will
1014
load non-polymorphically by default. Together, these two
1015
features allow a non-polymorphic concrete inheritance setup to
1016
use polymorphic loading on a per-query basis, since concrete
1017
setups are prone to many issues when used polymorphically in
1020
- dynamic_loader accepts a query_class= to customize the Query
1021
classes used for both the dynamic collection and the queries
1024
- query.order_by() accepts None which will remove any pending
1025
order_by state from the query, as well as cancel out any
1026
mapper/relation configured ordering. This is primarily useful
1027
for overriding the ordering specified on a dynamic_loader().
1031
- RowProxy objects can be used in place of dictionary arguments
1032
sent to connection.execute() and friends. [ticket:935]
1035
- Added a new description_encoding attribute on the dialect that
1036
is used for encoding the column name when processing the
1037
metadata. This usually defaults to utf-8.
1040
- Added in a new MSGenericBinary type. This maps to the Binary
1041
type so it can implement the specialized behavior of treating
1042
length specified types as fixed-width Binary types and
1043
non-length types as an unbound variable length Binary type.
1045
- Added in new types: MSVarBinary and MSImage. [ticket:1249]
1047
- Added in the MSReal, MSNText, MSSmallDateTime, MSTime,
1048
MSDateTimeOffset, and MSDateTime2 types
1051
- Table reflection now stores the actual DefaultClause value for
1052
the column. [ticket:1266]
1054
- bugfixes, behavioral changes
1057
- Exceptions raised during compile_mappers() are now preserved
1058
to provide "sticky behavior" - if a hasattr() call on a
1059
pre-compiled mapped attribute triggers a failing compile and
1060
suppresses the exception, subsequent compilation is blocked
1061
and the exception will be reiterated on the next compile()
1062
call. This issue occurs frequently when using declarative.
1064
- property.of_type() is now recognized on a single-table
1065
inheriting target, when used in the context of
1066
prop.of_type(..).any()/has(), as well as
1067
query.join(prop.of_type(...)).
1069
- query.join() raises an error when the target of the join
1070
doesn't match the property-based attribute - while it's
1071
unlikely anyone is doing this, the SQLAlchemy author was
1072
guilty of this particular loosey-goosey behavior.
1074
- Fixed bug when using weak_instance_map=False where modified
1075
events would not be intercepted for a flush(). [ticket:1272]
1077
- Fixed some deep "column correspondence" issues which could
1078
impact a Query made against a selectable containing multiple
1079
versions of the same table, as well as unions and similar
1080
which contained the same table columns in different column
1081
positions at different levels. [ticket:1268]
1083
- Custom comparator classes used in conjunction with
1084
column_property(), relation() etc. can define new comparison
1085
methods on the Comparator, which will become available via
1086
__getattr__() on the InstrumentedAttribute. In the case of
1087
synonym() or comparable_property(), attributes are resolved
1088
first on the user-defined descriptor, then on the user-defined
1091
- Added ScopedSession.is_active accessor. [ticket:976]
1093
- Can pass mapped attributes and column objects as keys to
1094
query.update({}). [ticket:1262]
1096
- Mapped attributes passed to the values() of an expression
1097
level insert() or update() will use the keys of the mapped
1098
columns, not that of the mapped attribute.
1100
- Corrected problem with Query.delete() and Query.update() not
1101
working properly with bind parameters. [ticket:1242]
1103
- Query.select_from(), from_statement() ensure that the given
1104
argument is a FromClause, or Text/Select/Union, respectively.
1106
- Query() can be passed a "composite" attribute as a column
1107
expression and it will be expanded. Somewhat related to
1110
- Query() is a little more robust when passed various column
1111
expressions such as strings, clauselists, text() constructs
1112
(which may mean it just raises an error more nicely).
1114
- first() works as expected with Query.from_statement().
1116
- Fixed bug introduced in 0.5rc4 involving eager loading not
1117
functioning for properties which were added to a mapper
1118
post-compile using add_property() or equivalent.
1120
- Fixed bug where many-to-many relation() with viewonly=True
1121
would not correctly reference the link between
1124
- Duplicate items in a list-based collection will be maintained
1125
when issuing INSERTs to a "secondary" table in a many-to-many
1126
relation. Assuming the m2m table has a unique or primary key
1127
constraint on it, this will raise the expected constraint
1128
violation instead of silently dropping the duplicate
1129
entries. Note that the old behavior remains for a one-to-many
1130
relation since collection entries in that case don't result in
1131
INSERT statements and SQLA doesn't manually police
1132
collections. [ticket:1232]
1134
- Query.add_column() can accept FromClause objects in the same
1135
manner as session.query() can.
1137
- Comparison of many-to-one relation to NULL is properly
1138
converted to IS NOT NULL based on not_().
1140
- Extra checks added to ensure explicit
1141
primaryjoin/secondaryjoin are ClauseElement instances, to
1142
prevent more confusing errors later on. [ticket:1087]
1144
- Improved mapper() check for non-class classes. [ticket:1236]
1146
- comparator_factory argument is now documented and supported by
1147
all MapperProperty types, including column_property(),
1148
relation(), backref(), and synonym() [ticket:5051].
1150
- Changed the name of PropertyLoader to RelationProperty, to be
1151
consistent with all the other names. PropertyLoader is still
1152
present as a synonym.
1154
- fixed "double iter()" call causing bus errors in shard API,
1155
removed errant result.close() left over from the 0.4
1156
version. [ticket:1099] [ticket:1228]
1158
- made Session.merge cascades not trigger autoflush. Fixes
1159
merged instances getting prematurely inserted with missing
1162
- Two fixes to help prevent out-of-band columns from being
1163
rendered in polymorphic_union inheritance scenarios (which
1164
then causes extra tables to be rendered in the FROM clause
1165
causing cartesian products):
1167
- improvements to "column adaption" for a->b->c inheritance
1168
situations to better locate columns that are related to
1169
one another via multiple levels of indirection, rather
1170
than rendering the non-adapted column.
1172
- the "polymorphic discriminator" column is only rendered
1173
for the actual mapper being queried against. The column
1174
won't be "pulled in" from a subclass or superclass mapper
1175
since it's not needed.
1177
- Fixed shard_id argument on ShardedSession.execute().
1181
- Columns can again contain percent signs within their
1182
names. [ticket:1256]
1184
- sqlalchemy.sql.expression.Function is now a public class. It
1185
can be subclassed to provide user-defined SQL functions in an
1186
imperative style, including with pre-established behaviors.
1187
The postgis.py example illustrates one usage of this.
1189
- PickleType now favors == comparison by default, if the
1190
incoming object (such as a dict) implements __eq__(). If the
1191
object does not implement __eq__() and mutable=True, a
1192
deprecation warning is raised.
1194
- Fixed the import weirdness in sqlalchemy.sql to not export
1195
__names__ [ticket:1215].
1197
- Using the same ForeignKey object repeatedly raises an error
1198
instead of silently failing later. [ticket:1238]
1200
- Added NotImplementedError for params() method on
1201
Insert/Update/Delete constructs. These items currently don't
1202
support this functionality, which also would be a little
1203
misleading compared to values().
1205
- Reflected foreign keys will properly locate their referenced
1206
column, even if the column was given a "key" attribute
1207
different from the reflected name. This is achieved via a new
1208
flag on ForeignKey/ForeignKeyConstraint called "link_to_name",
1209
if True means the given name is the referred-to column's name,
1210
not its assigned key. [ticket:650]
1212
- select() can accept a ClauseList as a column in the same way
1213
as a Table or other selectable and the interior expressions
1214
will be used as column elements. [ticket:1253]
1216
- the "passive" flag on session.is_modified() is correctly
1217
propagated to the attribute manager.
1219
- union() and union_all() will not whack any order_by() that has
1220
been applied to the select()s inside. If you union() a
1221
select() with order_by() (presumably to support LIMIT/OFFSET),
1222
you should also call self_group() on it to apply parenthesis.
1226
- Connection.invalidate() checks for closed status to avoid
1227
attribute errors. [ticket:1246]
1229
- NullPool supports reconnect on failure behavior.
1232
- Added a mutex for the initial pool creation when using
1233
pool.manage(dbapi). This prevents a minor case of "dogpile"
1234
behavior which would otherwise occur upon a heavy load
1235
startup. [ticket:799]
1237
- _execute_clauseelement() goes back to being a private method.
1238
Subclassing Connection is not needed now that ConnectionProxy
1242
- Tickets [ticket:1200] [ticket:1149].
1244
- Added note about create_session() defaults.
1246
- Added section about metadata.reflect().
1248
- Updated `TypeDecorator` section.
1250
- Rewrote the "threadlocal" strategy section of the docs due to
1251
recent confusion over this feature.
1253
- Removed badly out of date 'polymorphic_fetch' and
1254
'select_table' docs from inheritance, reworked the second half
1255
of "joined table inheritance".
1257
- Documented `comparator_factory` kwarg, added new doc section
1258
"Custom Comparators".
1261
- Refactored the Date/Time types. The ``smalldatetime`` data
1262
type no longer truncates to a date only, and will now be
1263
mapped to the MSSmallDateTime type. [ticket:1254]
1265
- Corrected an issue with Numerics to accept an int.
1267
- Mapped ``char_length`` to the ``LEN()`` function.
1269
- If an ``INSERT`` includes a subselect the ``INSERT`` is
1270
converted from an ``INSERT INTO VALUES`` construct to a
1271
``INSERT INTO SELECT`` construct.
1273
- If the column is part of a ``primary_key`` it will be ``NOT
1274
NULL`` since MSSQL doesn't allow ``NULL`` in primary_key
1277
- ``MSBinary`` now returns a ``BINARY`` instead of an
1278
``IMAGE``. This is a backwards incompatible change in that
1279
``BINARY`` is a fixed length data type whereas ``IMAGE`` is a
1280
variable length data type. [ticket:1249]
1282
- ``get_default_schema_name`` is now reflected from the database
1283
based on the user's default schema. This only works with MSSQL
1284
2005 and later. [ticket:1258]
1286
- Added collation support through the use of a new collation
1287
argument. This is supported on the following types: char,
1288
nchar, varchar, nvarchar, text, ntext. [ticket:1248]
1290
- Changes to the connection string parameters favor DSN as the
1291
default specification for pyodbc. See the mssql.py docstring
1292
for detailed usage instructions.
1294
- Added experimental support of savepoints. It currently does
1295
not work fully with sessions.
1297
- Support for three levels of column nullability: NULL, NOT
1298
NULL, and the database's configured default. The default
1299
Column configuration (nullable=True) will now generate NULL in
1300
the DDL. Previously no specification was emitted and the
1301
database default would take effect (usually NULL, but not
1302
always). To explicitly request the database default,
1303
configure columns with nullable=None and no specification will
1304
be emitted in DDL. This is backwards incompatible
1305
behavior. [ticket:1243]
1308
- "%" signs in text() constructs are automatically escaped to
1309
"%%". Because of the backwards incompatible nature of this
1310
change, a warning is emitted if '%%' is detected in the
1311
string. [ticket:1267]
1313
- Calling alias.execute() in conjunction with
1314
server_side_cursors won't raise AttributeError.
1316
- Added Index reflection support to PostgreSQL, using a great
1317
patch we long neglected, submitted by Ken
1318
Kuhlman. [ticket:714]
1321
- Adjusted the format of create_xid() to repair two-phase
1322
commit. We now have field reports of Oracle two-phase commit
1323
working properly with this change.
1325
- Added OracleNVarchar type, produces NVARCHAR2, and also
1326
subclasses Unicode so that convert_unicode=True by default.
1327
NVARCHAR2 reflects into this type automatically so these
1328
columns pass unicode on a reflected table with no explicit
1329
convert_unicode=True flags. [ticket:1233]
1331
- Fixed bug which was preventing out params of certain types
1332
from being received; thanks a ton to huddlej at wwu.edu !
1336
- "%" signs in text() constructs are automatically escaped to
1337
"%%". Because of the backwards incompatible nature of this
1338
change, a warning is emitted if '%%' is detected in the
1341
- Fixed bug in exception raise when FK columns not present
1342
during reflection. [ticket:1241]
1344
- Fixed bug involving reflection of a remote-schema table with a
1345
foreign key ref to another table in that schema.
1348
- The association proxy properties are make themselves available
1349
at the class level, e.g. MyClass.aproxy. Previously this
1353
- The full list of arguments accepted as string by backref()
1354
includes 'primaryjoin', 'secondaryjoin', 'secondary',
1355
'foreign_keys', 'remote_side', 'order_by'.
1361
- Query.count() has been enhanced to do the "right thing" in a
1362
wider variety of cases. It can now count multiple-entity
1363
queries, as well as column-based queries. Note that this means
1364
if you say query(A, B).count() without any joining criterion,
1365
it's going to count the cartesian product of A*B. Any query
1366
which is against column-based entities will automatically
1367
issue "SELECT count(1) FROM (SELECT...)" so that the real
1368
rowcount is returned, meaning a query such as
1369
query(func.count(A.name)).count() will return a value of one,
1370
since that query would return one row.
1372
- Lots of performance tuning. A rough guesstimate over various
1373
ORM operations places it 10% faster over 0.5.0rc3, 25-30% over
1376
- bugfixes and behavioral changes
1378
- global "propigate"->"propagate" change.
1381
- Adjustments to the enhanced garbage collection on
1382
InstanceState to better guard against errors due to lost
1385
- Query.get() returns a more informative error message when
1386
executed against multiple entities. [ticket:1220]
1388
- Restored NotImplementedError on Cls.relation.in_()
1389
[ticket:1140] [ticket:1221]
1391
- Fixed PendingDeprecationWarning involving order_by parameter
1392
on relation(). [ticket:1226]
1395
- Removed the 'properties' attribute of the Connection object,
1396
Connection.info should be used.
1398
- Restored "active rowcount" fetch before ResultProxy autocloses
1399
the cursor. This was removed in 0.5rc3.
1401
- Rearranged the `load_dialect_impl()` method in `TypeDecorator`
1402
such that it will take effect even if the user-defined
1403
`TypeDecorator` uses another `TypeDecorator` as its impl.
1406
- Added support for Currency type.
1408
- Functions were not return their result. [ticket:1017]
1410
- Corrected problem with joins. Access only support LEFT OUTER
1411
or INNER not just JOIN by itself. [ticket:1017]
1414
- Lots of cleanup and fixes to correct problems with limit and
1417
- Correct situation where subqueries as part of a binary
1418
expression need to be translated to use the IN and NOT IN
1421
- Fixed E Notation issue that prevented the ability to insert
1422
decimal values less than 1E-6. [ticket:1216]
1424
- Corrected problems with reflection when dealing with schemas,
1425
particularly when those schemas are the default
1426
schema. [ticket:1217]
1428
- Corrected problem with casting a zero length item to a
1429
varchar. It now correctly adjusts the CAST.
1432
- Can now use a custom "inherit_condition" in __mapper_args__
1433
when using declarative.
1435
- fixed string-based "remote_side", "order_by" and others not
1436
propagating correctly when used in backref().
1442
- Added two new hooks to SessionExtension: after_bulk_delete()
1443
and after_bulk_update(). after_bulk_delete() is called after
1444
a bulk delete() operation on a query. after_bulk_update() is
1445
called after a bulk update() operation on a query.
1448
- SQL compiler optimizations and complexity reduction. The call
1449
count for compiling a typical select() construct is 20% less
1452
- Dialects can now generate label names of adjustable
1453
length. Pass in the argument "label_length=<value>" to
1454
create_engine() to adjust how many characters max will be
1455
present in dynamically generated column labels, i.e.
1456
"somecolumn AS somelabel". Any value less than 6 will result
1457
in a label of minimal size, consisting of an underscore and a
1458
numeric counter. The compiler uses the value of
1459
dialect.max_identifier_length as a default. [ticket:1211]
1462
- Added a new extension sqlalchemy.ext.serializer. Provides
1463
Serializer/Deserializer "classes" which mirror
1464
Pickle/Unpickle, as well as dumps() and loads(). This
1465
serializer implements an "external object" pickler which keeps
1466
key context-sensitive objects, including engines, sessions,
1467
metadata, Tables/Columns, and mappers, outside of the pickle
1468
stream, and can later restore the pickle using any
1469
engine/metadata/session provider. This is used not for
1470
pickling regular object instances, which are pickleable
1471
without any special logic, but for pickling expression objects
1472
and full Query objects, such that all mapper/engine/session
1473
dependencies can be restored at unpickle time.
1476
- Wrote a docstring for Oracle dialect. Apparently that Ohloh
1477
"few source code comments" label is starting to sting :).
1479
- Removed FIRST_ROWS() optimize flag when using LIMIT/OFFSET,
1480
can be reenabled with optimize_limits=True create_engine()
1483
- bugfixes and behavioral changes
1485
- "not equals" comparisons of simple many-to-one relation to an
1486
instance will not drop into an EXISTS clause and will compare
1487
foreign key columns instead.
1489
- Removed not-really-working use cases of comparing a collection
1490
to an iterable. Use contains() to test for collection
1493
- Improved the behavior of aliased() objects such that they more
1494
accurately adapt the expressions generated, which helps
1495
particularly with self-referential comparisons. [ticket:1171]
1497
- Fixed bug involving primaryjoin/secondaryjoin conditions
1498
constructed from class-bound attributes (as often occurs when
1499
using declarative), which later would be inappropriately
1500
aliased by Query, particularly with the various EXISTS based
1503
- Fixed bug when using multiple query.join() with an
1504
aliased-bound descriptor which would lose the left alias.
1506
- Improved weakref identity map memory management to no longer
1507
require mutexing, resurrects garbage collected instance on a
1508
lazy basis for an InstanceState with pending changes.
1510
- InstanceState object now removes circular references to itself
1511
upon disposal to keep it outside of cyclic garbage collection.
1513
- relation() won't hide unrelated ForeignKey errors inside of
1514
the "please specify primaryjoin" message when determining join
1517
- Fixed bug in Query involving order_by() in conjunction with
1518
multiple aliases of the same class (will add tests in
1521
- When using Query.join() with an explicit clause for the ON
1522
clause, the clause will be aliased in terms of the left side
1523
of the join, allowing scenarios like query(Source).
1524
from_self().join((Dest, Source.id==Dest.source_id)) to work
1527
- polymorphic_union() function respects the "key" of each Column
1528
if they differ from the column's name.
1530
- Repaired support for "passive-deletes" on a many-to-one
1531
relation() with "delete" cascade. [ticket:1183]
1533
- Fixed bug in composite types which prevented a primary-key
1534
composite type from being mutated [ticket:1213].
1536
- Added more granularity to internal attribute access, such that
1537
cascade and flush operations will not initialize unloaded
1538
attributes and collections, leaving them intact for a
1539
lazy-load later on. Backref events still initialize attrbutes
1540
and collections for pending instances. [ticket:1202]
1543
- Simplified the check for ResultProxy "autoclose without
1544
results" to be based solely on presence of
1545
cursor.description. All the regexp-based guessing about
1546
statements returning rows has been removed [ticket:1212].
1548
- Direct execution of a union() construct will properly set up
1549
result-row processing. [ticket:1194]
1551
- The internal notion of an "OID" or "ROWID" column has been
1552
removed. It's basically not used by any dialect, and the
1553
possibility of its usage with psycopg2's cursor.lastrowid is
1554
basically gone now that INSERT..RETURNING is available.
1556
- Removed "default_order_by()" method on all FromClause objects.
1558
- Repaired the table.tometadata() method so that a passed-in
1559
schema argument is propagated to ForeignKey constructs.
1561
- Slightly changed behavior of IN operator for comparing to
1562
empty collections. Now results in inequality comparison
1563
against self. More portable, but breaks with stored procedures
1564
that aren't pure functions.
1567
- Setting the auto_convert_lobs to False on create_engine() will
1568
also instruct the OracleBinary type to return the cx_oracle
1569
LOB object unchanged.
1572
- Fixed foreign key reflection in the edge case where a Table's
1573
explicit schema= is the same as the schema (database) the
1574
connection is attached to.
1576
- No longer expects include_columns in table reflection to be
1580
- Fixed bug preventing declarative-bound "column" objects from
1581
being used in column_mapped_collection(). [ticket:1174]
1584
- util.flatten_iterator() func doesn't interpret strings with
1585
__iter__() methods as iterators, such as in pypy
1591
- Fixed bug involving read/write relation()s that contain
1592
literal or other non-column expressions within their
1593
primaryjoin condition equated to a foreign key column.
1595
- "non-batch" mode in mapper(), a feature which allows mapper
1596
extension methods to be called as each instance is
1597
updated/inserted, now honors the insert order of the objects
1600
- Fixed RLock-related bug in mapper which could deadlock upon
1601
reentrant mapper compile() calls, something that occurs when
1602
using declarative constructs inside of ForeignKey objects.
1604
- ScopedSession.query_property now accepts a query_cls factory,
1605
overriding the session's configured query_cls.
1607
- Fixed shared state bug interfering with ScopedSession.mapper's
1608
ability to apply default __init__ implementations on object
1611
- Fixed up slices on Query (i.e. query[x:y]) to work properly
1612
for zero length slices, slices with None on either end.
1615
- Added an example illustrating Celko's "nested sets" as a
1618
- contains_eager() with an alias argument works even when
1619
the alias is embedded in a SELECT, as when sent to the
1620
Query via query.select_from().
1622
- contains_eager() usage is now compatible with a Query that
1623
also contains a regular eager load and limit/offset, in that
1624
the columns are added to the Query-generated subquery.
1627
- session.execute() will execute a Sequence object passed to
1628
it (regression from 0.4).
1630
- Removed the "raiseerror" keyword argument from object_mapper()
1631
and class_mapper(). These functions raise in all cases
1632
if the given class/instance is not mapped.
1634
- Fixed session.transaction.commit() on a autocommit=False
1635
session not starting a new transaction.
1637
- Some adjustments to Session.identity_map's weak referencing
1638
behavior to reduce asynchronous GC side effects.
1640
- Adjustment to Session's post-flush accounting of newly
1641
"clean" objects to better protect against operating on
1642
objects as they're asynchronously gc'ed. [ticket:1182]
1645
- column.in_(someselect) can now be used as a columns-clause
1646
expression without the subquery bleeding into the FROM clause
1650
- Overhauled SQLite date/time bind/result processing to use
1651
regular expressions and format strings, rather than
1652
strptime/strftime, to generically support pre-1900 dates,
1653
dates with microseconds. [ticket:968]
1655
- String's (and Unicode's, UnicodeText's, etc.) convert_unicode
1656
logic disabled in the sqlite dialect, to adjust for pysqlite
1657
2.5.0's new requirement that only Python unicode objects are
1659
http://itsystementwicklung.de/pipermail/list-pysqlite/2008-March/000018.html
1662
- Temporary tables are now reflectable.
1665
- Oracle will detect string-based statements which contain
1666
comments at the front before a SELECT as SELECT statements.
1673
- Query now has delete() and update(values) methods. This allows
1674
to perform bulk deletes/updates with the Query object.
1676
- The RowTuple object returned by Query(*cols) now features
1677
keynames which prefer mapped attribute names over column keys,
1678
column keys over column names, i.e. Query(Class.foo,
1679
Class.bar) will have names "foo" and "bar" even if those are
1680
not the names of the underlying Column objects. Direct Column
1681
objects such as Query(table.c.col) will return the "key"
1682
attribute of the Column.
1684
- Added scalar() and value() methods to Query, each return a
1685
single scalar value. scalar() takes no arguments and is
1686
roughly equivalent to first()[0], value()
1687
takes a single column expression and is roughly equivalent to
1688
values(expr).next()[0].
1690
- Improved the determination of the FROM clause when placing SQL
1691
expressions in the query() list of entities. In particular
1692
scalar subqueries should not "leak" their inner FROM objects
1693
out into the enclosing query.
1695
- Joins along a relation() from a mapped class to a mapped
1696
subclass, where the mapped subclass is configured with single
1697
table inheritance, will include an IN clause which limits the
1698
subtypes of the joined class to those requested, within the ON
1699
clause of the join. This takes effect for eager load joins as
1700
well as query.join(). Note that in some scenarios the IN
1701
clause will appear in the WHERE clause of the query as well
1702
since this discrimination has multiple trigger points.
1704
- AttributeExtension has been refined such that the event
1705
is fired before the mutation actually occurs. Additionally,
1706
the append() and set() methods must now return the given value,
1707
which is used as the value to be used in the mutation operation.
1708
This allows creation of validating AttributeListeners which
1709
raise before the action actually occurs, and which can change
1710
the given value into something else before its used.
1712
- column_property(), composite_property(), and relation() now
1713
accept a single or list of AttributeExtensions using the
1714
"extension" keyword argument.
1716
- query.order_by().get() silently drops the "ORDER BY" from
1717
the query issued by GET but does not raise an exception.
1719
- Added a Validator AttributeExtension, as well as a
1720
@validates decorator which is used in a similar fashion
1721
as @reconstructor, and marks a method as validating
1722
one or more mapped attributes.
1724
- class.someprop.in_() raises NotImplementedError pending the
1725
implementation of "in_" for relation [ticket:1140]
1727
- Fixed primary key update for many-to-many collections where
1728
the collection had not been loaded yet [ticket:1127]
1730
- Fixed bug whereby deferred() columns with a group in conjunction
1731
with an otherwise unrelated synonym() would produce
1732
an AttributeError during deferred load.
1734
- The before_flush() hook on SessionExtension takes place before
1735
the list of new/dirty/deleted is calculated for the final
1736
time, allowing routines within before_flush() to further
1737
change the state of the Session before the flush proceeds.
1740
- The "extension" argument to Session and others can now
1741
optionally be a list, supporting events sent to multiple
1742
SessionExtension instances. Session places SessionExtensions
1743
in Session.extensions.
1745
- Reentrant calls to flush() raise an error. This also serves
1746
as a rudimentary, but not foolproof, check against concurrent
1747
calls to Session.flush().
1749
- Improved the behavior of query.join() when joining to
1750
joined-table inheritance subclasses, using explicit join
1751
criteria (i.e. not on a relation).
1753
- @orm.attributes.reconstitute and
1754
MapperExtension.reconstitute have been renamed to
1755
@orm.reconstructor and MapperExtension.reconstruct_instance
1757
- Fixed @reconstructor hook for subclasses which inherit from a
1758
base class. [ticket:1129]
1760
- The composite() property type now supports a
1761
__set_composite_values__() method on the composite class which
1762
is required if the class represents state using attribute
1763
names other than the column's keynames; default-generated
1764
values now get populated properly upon flush. Also,
1765
composites with attributes set to None compare correctly.
1768
- The 3-tuple of iterables returned by attributes.get_history()
1769
may now be a mix of lists and tuples. (Previously members
1772
- Fixed bug whereby changing a primary key attribute on an
1773
entity where the attribute's previous value had been expired
1774
would produce an error upon flush(). [ticket:1151]
1776
- Fixed custom instrumentation bug whereby get_instance_dict()
1777
was not called for newly constructed instances not loaded
1780
- Session.delete() adds the given object to the session if
1781
not already present. This was a regression bug from 0.4.
1784
- The `echo_uow` flag on `Session` is deprecated, and unit-of-work
1785
logging is now application-level only, not per-session level.
1787
- Removed conflicting `contains()` operator from
1788
`InstrumentedAttribute` which didn't accept `escape` kwaarg
1792
- Fixed bug whereby mapper couldn't initialize if a composite
1793
primary key referenced another table that was not defined
1796
- Fixed exception throw which would occur when string-based
1797
primaryjoin condition was used in conjunction with backref.
1800
- Added "sorted_tables" accessor to MetaData, which returns
1801
Table objects sorted in order of dependency as a list.
1802
This deprecates the MetaData.table_iterator() method.
1803
The "reverse=False" keyword argument has also been
1804
removed from util.sort_tables(); use the Python
1805
'reversed' function to reverse the results.
1808
- The 'length' argument to all Numeric types has been renamed
1809
to 'scale'. 'length' is deprecated and is still accepted
1812
- Dropped 0.3-compatibility for user defined types
1813
(convert_result_value, convert_bind_param).
1816
- Temporarily rolled back the "ORDER BY" enhancement from
1817
[ticket:1068]. This feature is on hold pending further
1820
- The exists() construct won't "export" its contained list
1821
of elements as FROM clauses, allowing them to be used more
1822
effectively in the columns clause of a SELECT.
1824
- and_() and or_() now generate a ColumnElement, allowing
1825
boolean expressions as result columns, i.e.
1826
select([and_(1, 0)]). [ticket:798]
1828
- Bind params now subclass ColumnElement which allows them to be
1829
selectable by orm.query (they already had most ColumnElement
1832
- Added select_from() method to exists() construct, which becomes
1833
more and more compatible with a regular select().
1835
- Added func.min(), func.max(), func.sum() as "generic functions",
1836
which basically allows for their return type to be determined
1837
automatically. Helps with dates on SQLite, decimal types,
1838
others. [ticket:1160]
1840
- added decimal.Decimal as an "auto-detect" type; bind parameters
1841
and generic functions will set their type to Numeric when a
1845
- The 'length' argument to MSInteger, MSBigInteger, MSTinyInteger,
1846
MSSmallInteger and MSYear has been renamed to 'display_width'.
1848
- Added MSMediumInteger type [ticket:1146].
1850
- the function func.utc_timestamp() compiles to UTC_TIMESTAMP, without
1851
the parenthesis, which seem to get in the way when using in
1852
conjunction with executemany().
1855
- limit/offset no longer uses ROW NUMBER OVER to limit rows,
1856
and instead uses subqueries in conjunction with a special
1857
Oracle optimization comment. Allows LIMIT/OFFSET to work
1858
in conjunction with DISTINCT. [ticket:536]
1859
- has_sequence() now takes the current "schema" argument into
1860
account [ticket:1155]
1861
- added BFILE to reflected type names [ticket:1121]
1866
0.5beta3 includes all bugfixes listed under release "0.4.7".
1869
- The "entity_name" feature of SQLAlchemy mappers has been
1870
removed. For rationale, see http://tinyurl.com/6nm2ne
1872
- the "autoexpire" flag on Session, sessionmaker(), and
1873
scoped_session() has been renamed to "expire_on_commit". It
1874
does not affect the expiration behavior of rollback().
1876
- fixed endless loop bug which could occur within a mapper's
1877
deferred load of inherited attributes.
1879
- a legacy-support flag "_enable_transaction_accounting" flag
1880
added to Session which when False, disables all
1881
transaction-level object accounting, including expire on
1882
rollback, expire on commit, new/deleted list maintenance, and
1885
- The 'cascade' parameter to relation() accepts None as a value,
1886
which is equivalent to no cascades.
1888
- A critical fix to dynamic relations allows the "modified"
1889
history to be properly cleared after a flush().
1891
- user-defined @properties on a class are detected and left in
1892
place during mapper initialization. This means that a
1893
table-bound column of the same name will not be mapped at all
1894
if a @property is in the way (and the column is not remapped
1895
to a different name), nor will an instrumented attribute from
1896
an inherited class be applied. The same rules apply for names
1897
excluded using the include_properties/exclude_properties
1900
- Added a new SessionExtension hook called after_attach(). This
1901
is called at the point of attachment for objects via add(),
1902
add_all(), delete(), and merge().
1904
- A mapper which inherits from another, when inheriting the
1905
columns of its inherited mapper, will use any reassigned
1906
property names specified in that inheriting mapper.
1907
Previously, if "Base" had reassigned "base_id" to the name
1908
"id", "SubBase(Base)" would still get an attribute called
1909
"base_id". This could be worked around by explicitly stating
1910
the column in each submapper as well but this is fairly
1911
unworkable and also impossible when using declarative
1914
- Fixed a series of potential race conditions in Session whereby
1915
asynchronous GC could remove unmodified, no longer referenced
1916
items from the session as they were present in a list of items
1917
to be processed, typically during session.expunge_all() and
1920
- Some improvements to the _CompileOnAttr mechanism which should
1921
reduce the probability of "Attribute x was not replaced during
1922
compile" warnings. (this generally applies to SQLA hackers,
1925
- Fixed bug whereby the "unsaved, pending instance" FlushError
1926
raised for a pending orphan would not take superclass mappers
1927
into account when generating the list of relations responsible
1931
- func.count() with no arguments renders as COUNT(*), equivalent
1932
to func.count(text('*')).
1934
- simple label names in ORDER BY expressions render as
1935
themselves, and not as a re-statement of their corresponding
1936
expression. This feature is currently enabled only for
1937
SQLite, MySQL, and PostgreSQL. It can be enabled on other
1938
dialects as each is shown to support this
1939
behavior. [ticket:1068]
1942
- Class-bound attributes sent as arguments to relation()'s
1943
remote_side and foreign_keys parameters are now accepted,
1944
allowing them to be used with declarative. Additionally fixed
1945
bugs involving order_by being specified as a class-bound
1946
attribute in conjunction with eager loading.
1948
- declarative initialization of Columns adjusted so that
1949
non-renamed columns initialize in the same way as a non
1950
declarative mapper. This allows an inheriting mapper to set
1951
up its same-named "id" columns in particular such that the
1952
parent "id" column is favored over the child column, reducing
1953
database round trips when this value is requested.
1956
- Quoting of MSEnum values for use in CREATE TABLE is now
1957
optional & will be quoted on demand as required. (Quoting was
1958
always optional for use with existing tables.) [ticket:1110]
1963
0.5beta2 includes some of the bugfixes listed under release "0.4.7".
1966
- In addition to expired attributes, deferred attributes also
1967
load if their data is present in the result set. [ticket:870]
1969
- session.refresh() raises an informative error message if the
1970
list of attributes does not include any column-based
1973
- query() raises an informative error message if no columns or
1974
mappers are specified.
1976
- lazy loaders now trigger autoflush before proceeding. This
1977
allows expire() of a collection or scalar relation to function
1978
properly in the context of autoflush.
1980
- column_property() attributes which represent SQL expressions
1981
or columns that are not present in the mapped tables (such as
1982
those from views) are automatically expired after an INSERT or
1983
UPDATE, assuming they have not been locally modified, so that
1984
they are refreshed with the most recent data upon access.
1987
- Fixed explicit, self-referential joins between two
1988
joined-table inheritance mappers when using query.join(cls,
1989
aliased=True). [ticket:1082]
1991
- Fixed query.join() when used in conjunction with a
1992
columns-only clause and an SQL-expression ON clause in the
1995
- The "allow_column_override" flag from mapper() has been
1996
removed. This flag is virtually always misunderstood. Its
1997
specific functionality is available via the
1998
include_properties/exclude_properties mapper arguments.
2000
- Repaired `__str__()` method on Query. [ticket:1066]
2002
- Session.bind gets used as a default even when table/mapper
2003
specific binds are defined.
2006
- Added prefixes option to `Table` that accepts a list of
2007
strings to insert after CREATE in the CREATE TABLE statement.
2010
- Unicode, UnicodeText types now set "assert_unicode" and
2011
"convert_unicode" by default, but accept overriding
2012
**kwargs for these values.
2015
- Added new match() operator that performs a full-text search.
2016
Supported on PostgreSQL, SQLite, MySQL, MS-SQL, and Oracle
2020
- Modified SQLite's representation of "microseconds" to match
2021
the output of str(somedatetime), i.e. in that the microseconds
2022
are represented as fractional seconds in string format. This
2023
makes SQLA's SQLite date type compatible with datetimes that
2024
were saved directly using Pysqlite (which just calls str()).
2025
Note that this is incompatible with the existing microseconds
2026
values in a SQLA 0.4 generated SQLite database file.
2028
To get the old behavior globally:
2030
from sqlalchemy.databases.sqlite import DateTimeMixin
2031
DateTimeMixin.__legacy_microseconds__ = True
2033
To get the behavior on individual DateTime types:
2035
t = sqlite.SLDateTime()
2036
t.__legacy_microseconds__ = True
2038
Then use "t" as the type on the Column.
2042
- SQLite Date, DateTime, and Time types only accept Python
2043
datetime objects now, not strings. If you'd like to format
2044
dates as strings yourself with SQLite, use a String type. If
2045
you'd like them to return datetime objects anyway despite
2046
their accepting strings as input, make a TypeDecorator around
2047
String - SQLA doesn't encourage this pattern.
2050
- Declarative supports a __table_args__ class variable, which is
2051
either a dictionary, or tuple of the form (arg1, arg2, ...,
2052
{kwarg1:value, ...}) which contains positional + kw arguments
2053
to be passed to the Table constructor. [ticket:1096]
2058
An ongoing document describing the changes from 0.4 to 0.5 is at:
2059
http://www.sqlalchemy.org/trac/wiki/05Migration
2061
For this section, the list below primarily lists changes within the
2062
attribute instrumentation API and is only a small fraction of the
2063
full changes between 0.4/0.5.
2066
- The "__init__" trigger/decorator added by mapper now attempts
2067
to exactly mirror the argument signature of the original
2068
__init__. The pass-through for '_sa_session' is no longer
2069
implicit- you must allow for this keyword argument in your
2072
- ClassState is renamed to ClassManager.
2074
- Classes may supply their own InstrumentationManager by
2075
providing a __sa_instrumentation_manager__ property.
2077
- Custom instrumentation may use any mechanism to associate a
2078
ClassManager with a class and an InstanceState with an
2079
instance. Attributes on those objects are still the default
2080
association mechanism used by SQLAlchemy's native
2083
- Moved entity_name, _sa_session_id, and _instance_key from the
2084
instance object to the instance state. These values are still
2085
available in the old way, which is now deprecated, using
2086
descriptors attached to the class. A deprecation warning will
2087
be issued when accessed.
2089
- The _prepare_instrumentation alias for prepare_instrumentation
2092
- sqlalchemy.exceptions has been renamed to sqlalchemy.exc. The
2093
module may be imported under either name.
2095
- ORM-related exceptions are now defined in sqlalchemy.orm.exc.
2096
ConcurrentModificationError, FlushError, and
2097
UnmappedColumnError compatibility aliases are installed in
2098
sqlalchemy.exc during the import of sqlalchemy.orm.
2100
- sqlalchemy.logging has been renamed to sqlalchemy.log.
2102
- The transitional sqlalchemy.log.SADeprecationWarning alias for
2103
the warning's definition in sqlalchemy.exc has been removed.
2105
- exc.AssertionError has been removed and usage replaced with
2106
Python's built-in AssertionError.
2108
- The behavior of MapperExtensions attached to multiple,
2109
entity_name= primary mappers for a single class has been
2110
altered. The first mapper() defined for a class is the only
2111
mapper eligible for the MapperExtension 'instrument_class',
2112
'init_instance' and 'init_failed' events. This is backwards
2113
incompatible; previously the extensions of last mapper defined
2114
would receive these events.
2117
- Added support for returning values from inserts (2.0+ only),
2118
updates and deletes (2.1+ only).
2123
- The contains() operator when used with many-to-many
2124
will alias() the secondary (association) table so
2125
that multiple contains() calls will not conflict
2126
with each other [ticket:1058]
2128
- fixed bug preventing merge() from functioning in
2129
conjunction with a comparable_property()
2131
- the enable_typechecks=False setting on relation()
2132
now only allows subtypes with inheriting mappers.
2133
Totally unrelated types, or subtypes not set up with
2134
mapper inheritance against the target mapper are
2137
- Added is_active flag to Sessions to detect when
2138
a transaction is in progress [ticket:976]. This
2139
flag is always True with a "transactional"
2140
(in 0.5 a non-"autocommit") Session.
2143
- Fixed bug when calling select([literal('foo')])
2144
or select([bindparam('foo')]).
2147
- create_all(), drop_all(), create(), drop() all raise
2148
an error if the table name or schema name contains
2149
more characters than that dialect's configured
2150
character limit. Some DB's can handle too-long
2151
table names during usage, and SQLA can handle this
2152
as well. But various reflection/
2153
checkfirst-during-create scenarios fail since we are
2154
looking for the name within the DB's catalog tables.
2157
- The index name generated when you say "index=True"
2158
on a Column is truncated to the length appropriate
2159
for the dialect. Additionally, an Index with a too-
2160
long name cannot be explicitly dropped with
2161
Index.drop(), similar to [ticket:571].
2165
- Repaired server_side_cursors to properly detect
2168
- Added PGCidr type. [ticket:1092]
2171
- Added 'CALL' to the list of SQL keywords which return
2175
- Oracle get_default_schema_name() "normalizes" the name
2176
before returning, meaning it returns a lower-case name
2177
when the identifier is detected as case insensitive.
2179
- creating/dropping tables takes schema name into account
2180
when searching for the existing table, so that tables
2181
in other owner namespaces with the same name do not
2182
conflict [ticket:709]
2184
- Cursors now have "arraysize" set to 50 by default on
2185
them, the value of which is configurable using the
2186
"arraysize" argument to create_engine() with the
2187
Oracle dialect. This to account for cx_oracle's default
2188
setting of "1", which has the effect of many round trips
2189
being sent to Oracle. This actually works well in
2190
conjunction with BLOB/CLOB-bound cursors, of which
2191
there are any number available but only for the life of
2192
that row request (so BufferedColumnRow is still needed,
2193
but less so). [ticket:1062]
2196
- add SLFloat type, which matches the SQLite REAL
2197
type affinity. Previously, only SLNumeric was provided
2198
which fulfills NUMERIC affinity, but that's not the
2204
- Fix to the recent relation() refactoring which fixes
2205
exotic viewonly relations which join between local and
2206
remote table multiple times, with a common column shared
2209
- Also re-established viewonly relation() configurations
2210
that join across multiple tables.
2212
- Added experimental relation() flag to help with
2213
primaryjoins across functions, etc.,
2214
_local_remote_pairs=[tuples]. This complements a complex
2215
primaryjoin condition allowing you to provide the
2216
individual column pairs which comprise the relation's
2217
local and remote sides. Also improved lazy load SQL
2218
generation to handle placing bind params inside of
2219
functions and other expressions. (partial progress
2220
towards [ticket:610])
2222
- repaired single table inheritance such that you
2223
can single-table inherit from a joined-table inherting
2224
mapper without issue [ticket:1036].
2226
- Fixed "concatenate tuple" bug which could occur with
2227
Query.order_by() if clause adaption had taken place.
2230
- Removed ancient assertion that mapped selectables require
2231
"alias names" - the mapper creates its own alias now if
2232
none is present. Though in this case you need to use the
2233
class, not the mapped selectable, as the source of column
2234
attributes - so a warning is still issued.
2236
- fixes to the "exists" function involving inheritance (any(),
2237
has(), ~contains()); the full target join will be rendered
2238
into the EXISTS clause for relations that link to subclasses.
2240
- restored usage of append_result() extension method for primary
2241
query rows, when the extension is present and only a single-
2242
entity result is being returned.
2244
- Also re-established viewonly relation() configurations that
2245
join across multiple tables.
2247
- removed ancient assertion that mapped selectables require
2248
"alias names" - the mapper creates its own alias now if
2249
none is present. Though in this case you need to use
2250
the class, not the mapped selectable, as the source of
2251
column attributes - so a warning is still issued.
2253
- refined mapper._save_obj() which was unnecessarily calling
2254
__ne__() on scalar values during flush [ticket:1015]
2256
- added a feature to eager loading whereby subqueries set
2257
as column_property() with explicit label names (which is not
2258
necessary, btw) will have the label anonymized when
2259
the instance is part of the eager join, to prevent
2260
conflicts with a subquery or column of the same name
2261
on the parent object. [ticket:1019]
2263
- set-based collections |=, -=, ^= and &= are stricter about
2264
their operands and only operate on sets, frozensets or
2265
subclasses of the collection type. Previously, they would
2266
accept any duck-typed set.
2268
- added an example dynamic_dict/dynamic_dict.py, illustrating
2269
a simple way to place dictionary behavior on top of
2272
- declarative extension
2273
- Joined table inheritance mappers use a slightly relaxed
2274
function to create the "inherit condition" to the parent
2275
table, so that other foreign keys to not-yet-declared
2276
Table objects don't trigger an error.
2278
- fixed reentrant mapper compile hang when
2279
a declared attribute is used within ForeignKey,
2280
ie. ForeignKey(MyOtherClass.someattribute)
2283
- Added COLLATE support via the .collate(<collation>)
2284
expression operator and collate(<expr>, <collation>) sql
2287
- Fixed bug with union() when applied to non-Table connected
2290
- improved behavior of text() expressions when used as
2291
FROM clauses, such as select().select_from(text("sometext"))
2294
- Column.copy() respects the value of "autoincrement",
2295
fixes usage with Migrate [ticket:1021]
2298
- Pool listeners can now be provided as a dictionary of
2299
callables or a (possibly partial) duck-type of
2300
PoolListener, your choice.
2302
- added "rollback_returned" option to Pool which will
2303
disable the rollback() issued when connections are
2304
returned. This flag is only safe to use with a database
2305
which does not support transactions (i.e. MySQL/MyISAM).
2308
- set-based association proxies |=, -=, ^= and &= are
2309
stricter about their operands and only operate on sets,
2310
frozensets or other association proxies. Previously, they
2311
would accept any duck-typed set.
2314
- Added "odbc_autotranslate" parameter to engine / dburi
2315
parameters. Any given string will be passed through to the
2316
ODBC connection string as:
2318
"AutoTranslate=%s" % odbc_autotranslate
2322
- Added "odbc_options" parameter to engine / dburi
2323
parameters. The given string is simply appended to the
2324
SQLAlchemy-generated odbc connection string.
2326
This should obviate the need of adding a myriad of ODBC
2327
options in the future.
2330
- Handle the "SUBSTRING(:string FROM :start FOR :length)"
2336
- A small change in behavior to session.merge() - existing
2337
objects are checked for based on primary key attributes, not
2338
necessarily _instance_key. So the widely requested
2344
will in fact load MyObject with id #1 from the database if
2345
present, is now available. merge() still copies the state
2346
of the given object to the persistent one, so an example
2347
like the above would typically have copied "None" from all
2348
attributes of "x" onto the persistent copy. These can be
2349
reverted using session.expire(x).
2351
- Also fixed behavior in merge() whereby collection elements
2352
present on the destination but not the merged collection
2353
were not being removed from the destination.
2355
- Added a more aggressive check for "uncompiled mappers",
2356
helps particularly with declarative layer [ticket:995]
2358
- The methodology behind "primaryjoin"/"secondaryjoin" has
2359
been refactored. Behavior should be slightly more
2360
intelligent, primarily in terms of error messages which
2361
have been pared down to be more readable. In a slight
2362
number of scenarios it can better resolve the correct
2363
foreign key than before.
2365
- Added comparable_property(), adds query Comparator
2366
behavior to regular, unmanaged Python properties
2368
- the functionality of query.with_polymorphic() has
2369
been added to mapper() as a configuration option.
2371
It's set via several forms:
2372
with_polymorphic='*'
2373
with_polymorphic=[mappers]
2374
with_polymorphic=('*', selectable)
2375
with_polymorphic=([mappers], selectable)
2377
This controls the default polymorphic loading strategy
2378
for inherited mappers. When a selectable is not given,
2379
outer joins are created for all joined-table inheriting
2380
mappers requested. Note that the auto-create of joins
2381
is not compatible with concrete table inheritance.
2383
The existing select_table flag on mapper() is now
2384
deprecated and is synonymous with
2385
with_polymorphic('*', select_table). Note that the
2386
underlying "guts" of select_table have been
2387
completely removed and replaced with the newer,
2388
more flexible approach.
2390
The new approach also automatically allows eager loads
2391
to work for subclasses, if they are present, for
2393
sess.query(Company).options(
2395
[Company.employees.of_type(Engineer), 'machines']
2397
to load Company objects, their employees, and the
2398
'machines' collection of employees who happen to be
2399
Engineers. A "with_polymorphic" Query option should be
2400
introduced soon as well which would allow per-Query
2401
control of with_polymorphic() on relations.
2403
- added two "experimental" features to Query,
2404
"experimental" in that their specific name/behavior
2405
is not carved in stone just yet: _values() and
2406
_from_self(). We'd like feedback on these.
2408
- _values(*columns) is given a list of column
2409
expressions, and returns a new Query that only
2410
returns those columns. When evaluated, the return
2411
value is a list of tuples just like when using
2412
add_column() or add_entity(), the only difference is
2413
that "entity zero", i.e. the mapped class, is not
2414
included in the results. This means it finally makes
2415
sense to use group_by() and having() on Query, which
2416
have been sitting around uselessly until now.
2418
A future change to this method may include that its
2419
ability to join, filter and allow other options not
2420
related to a "resultset" are removed, so the feedback
2421
we're looking for is how people want to use
2422
_values()...i.e. at the very end, or do people prefer
2423
to continue generating after it's called.
2425
- _from_self() compiles the SELECT statement for the
2426
Query (minus any eager loaders), and returns a new
2427
Query that selects from that SELECT. So basically you
2428
can query from a Query without needing to extract the
2429
SELECT statement manually. This gives meaning to
2430
operations like query[3:5]._from_self().filter(some
2431
criterion). There's not much controversial here
2432
except that you can quickly create highly nested
2433
queries that are less efficient, and we want feedback
2434
on the naming choice.
2436
- query.order_by() and query.group_by() will accept
2437
multiple arguments using *args (like select()
2440
- Added some convenience descriptors to Query:
2441
query.statement returns the full SELECT construct,
2442
query.whereclause returns just the WHERE part of the
2445
- Fixed/covered case when using a False/0 value as a
2446
polymorphic discriminator.
2448
- Fixed bug which was preventing synonym() attributes from
2449
being used with inheritance
2451
- Fixed SQL function truncation of trailing underscores
2454
- When attributes are expired on a pending instance, an
2455
error will not be raised when the "refresh" action is
2456
triggered and no result is found.
2458
- Session.execute can now find binds from metadata
2460
- Adjusted the definition of "self-referential" to be any
2461
two mappers with a common parent (this affects whether or
2462
not aliased=True is required when joining with Query).
2464
- Made some fixes to the "from_joinpoint" argument to
2465
query.join() so that if the previous join was aliased and
2466
this one isn't, the join still happens successfully.
2468
- Assorted "cascade deletes" fixes:
2469
- Fixed "cascade delete" operation of dynamic relations,
2470
which had only been implemented for foreign-key
2471
nulling behavior in 0.4.2 and not actual cascading
2472
deletes [ticket:895]
2474
- Delete cascade without delete-orphan cascade on a
2475
many-to-one will not delete orphans which were
2476
disconnected from the parent before session.delete()
2477
is called on the parent (one-to-many already had
2480
- Delete cascade with delete-orphan will delete orphans
2481
whether or not it remains attached to its also-deleted
2484
- delete-orphan casacde is properly detected on relations
2485
that are present on superclasses when using inheritance.
2487
- Fixed order_by calculation in Query to properly alias
2488
mapper-config'ed order_by when using select_from()
2490
- Refactored the diffing logic that kicks in when replacing
2491
one collection with another into collections.bulk_replace,
2492
useful to anyone building multi-level collections.
2494
- Cascade traversal algorithm converted from recursive to
2495
iterative to support deep object graphs.
2498
- schema-qualified tables now will place the schemaname
2499
ahead of the tablename in all column expressions as well
2500
as when generating column labels. This prevents cross-
2501
schema name collisions in all cases [ticket:999]
2503
- can now allow selects which correlate all FROM clauses
2504
and have no FROM themselves. These are typically
2505
used in a scalar context, i.e. SELECT x, (SELECT x WHERE y)
2506
FROM table. Requires explicit correlate() call.
2508
- 'name' is no longer a required constructor argument for
2509
Column(). It (and .key) may now be deferred until the
2510
column is added to a Table.
2512
- like(), ilike(), contains(), startswith(), endswith() take
2513
an optional keyword argument "escape=<somestring>", which
2514
is set as the escape character using the syntax "x LIKE y
2515
ESCAPE '<somestring>'" [ticket:993], [ticket:791]
2517
- random() is now a generic sql function and will compile to
2518
the database's random implementation, if any.
2520
- update().values() and insert().values() take keyword
2523
- Fixed an issue in select() regarding its generation of
2524
FROM clauses, in rare circumstances two clauses could be
2525
produced when one was intended to cancel out the other.
2526
Some ORM queries with lots of eager loads might have seen
2529
- The case() function now also takes a dictionary as its
2530
whens parameter. It also interprets the "THEN"
2531
expressions as values by default, meaning case([(x==y,
2532
"foo")]) will interpret "foo" as a bound value, not a SQL
2533
expression. use text(expr) for literal SQL expressions in
2534
this case. For the criterion itself, these may be literal
2535
strings only if the "value" keyword is present, otherwise
2536
SA will force explicit usage of either text() or
2540
- The "owner" keyword on Table is now deprecated, and is
2541
exactly synonymous with the "schema" keyword. Tables can
2542
now be reflected with alternate "owner" attributes,
2543
explicitly stated on the Table object or not using
2546
- All of the "magic" searching for synonyms, DBLINKs etc.
2547
during table reflection are disabled by default unless you
2548
specify "oracle_resolve_synonyms=True" on the Table
2549
object. Resolving synonyms necessarily leads to some
2550
messy guessing which we'd rather leave off by default.
2551
When the flag is set, tables and related tables will be
2552
resolved against synonyms in all cases, meaning if a
2553
synonym exists for a particular table, reflection will use
2554
it when reflecting related tables. This is stickier
2555
behavior than before which is why it's off by default.
2557
- declarative extension
2558
- The "synonym" function is now directly usable with
2559
"declarative". Pass in the decorated property using the
2560
"descriptor" keyword argument, e.g.: somekey =
2561
synonym('_somekey', descriptor=property(g, s))
2563
- The "deferred" function is usable with "declarative".
2564
Simplest usage is to declare deferred and Column together,
2565
e.g.: data = deferred(Column(Text))
2567
- Declarative also gained @synonym_for(...) and
2568
@comparable_using(...), front-ends for synonym and
2569
comparable_property.
2571
- Improvements to mapper compilation when using declarative;
2572
already-compiled mappers will still trigger compiles of
2573
other uncompiled mappers when used [ticket:995]
2575
- Declarative will complete setup for Columns lacking names,
2576
allows a more DRY syntax.
2579
__tablename__ = 'foos'
2580
id = Column(Integer, primary_key=True)
2582
- inheritance in declarative can be disabled when sending
2583
"inherits=None" to __mapper_args__.
2585
- declarative_base() takes optional kwarg "mapper", which
2586
is any callable/class/method that produces a mapper,
2587
such as declarative_base(mapper=scopedsession.mapper).
2588
This property can also be set on individual declarative
2589
classes using the "__mapper_cls__" property.
2592
- Got PG server side cursors back into shape, added fixed
2593
unit tests as part of the default test suite. Added
2594
better uniqueness to the cursor ID [ticket:1001]
2597
- The "owner" keyword on Table is now deprecated, and is
2598
exactly synonymous with the "schema" keyword. Tables can
2599
now be reflected with alternate "owner" attributes,
2600
explicitly stated on the Table object or not using
2603
- All of the "magic" searching for synonyms, DBLINKs etc.
2604
during table reflection are disabled by default unless you
2605
specify "oracle_resolve_synonyms=True" on the Table
2606
object. Resolving synonyms necessarily leads to some
2607
messy guessing which we'd rather leave off by default.
2608
When the flag is set, tables and related tables will be
2609
resolved against synonyms in all cases, meaning if a
2610
synonym exists for a particular table, reflection will use
2611
it when reflecting related tables. This is stickier
2612
behavior than before which is why it's off by default.
2615
- Reflected tables will now automatically load other tables
2616
which are referenced by Foreign keys in the auto-loaded
2617
table, [ticket:979].
2619
- Added executemany check to skip identity fetch, [ticket:916].
2621
- Added stubs for small date type, [ticket:884]
2623
- Added a new 'driver' keyword parameter for the pyodbc dialect.
2624
Will substitute into the ODBC connection string if given,
2625
defaults to 'SQL Server'.
2627
- Added a new 'max_identifier_length' keyword parameter for
2630
- Improvements to pyodbc + Unix. If you couldn't get that
2631
combination to work before, please try again.
2634
- The connection.info keys the dialect uses to cache server
2635
settings have changed and are now namespaced.
2640
- Can again create aliases of selects against textual FROM
2641
clauses, [ticket:975]
2643
- The value of a bindparam() can be a callable, in which
2644
case it's evaluated at statement execution time to get the
2647
- Added exception wrapping/reconnect support to result set
2648
fetching. Reconnect works for those databases that raise
2649
a catchable data error during results (i.e. doesn't work
2650
on MySQL) [ticket:978]
2652
- Implemented two-phase API for "threadlocal" engine, via
2653
engine.begin_twophase(), engine.prepare() [ticket:936]
2655
- Fixed bug which was preventing UNIONS from being
2656
cloneable, [ticket:986]
2658
- Added "bind" keyword argument to insert(), update(),
2659
delete() and DDL(). The .bind property is now assignable
2660
on those statements as well as on select().
2662
- Insert statements can now be compiled with extra "prefix"
2663
words between INSERT and INTO, for vendor extensions like
2664
MySQL's INSERT IGNORE INTO table.
2667
- any(), has(), contains(), ~contains(), attribute level ==
2668
and != now work properly with self-referential relations -
2669
the clause inside the EXISTS is aliased on the "remote"
2670
side to distinguish it from the parent table. This
2671
applies to single table self-referential as well as
2672
inheritance-based self-referential.
2674
- Repaired behavior of == and != operators at the relation()
2675
level when compared against NULL for one-to-one relations
2678
- Fixed bug whereby session.expire() attributes were not
2679
loading on an polymorphically-mapped instance mapped by a
2680
select_table mapper.
2682
- Added query.with_polymorphic() - specifies a list of
2683
classes which descend from the base class, which will be
2684
added to the FROM clause of the query. Allows subclasses
2685
to be used within filter() criterion as well as eagerly
2686
loads the attributes of those subclasses.
2688
- Your cries have been heard: removing a pending item from
2689
an attribute or collection with delete-orphan expunges the
2690
item from the session; no FlushError is raised. Note that
2691
if you session.save()'ed the pending item explicitly, the
2692
attribute/collection removal still knocks it out.
2694
- session.refresh() and session.expire() raise an error when
2695
called on instances which are not persistent within the
2698
- Fixed potential generative bug when the same Query was
2699
used to generate multiple Query objects using join().
2701
- Fixed bug which was introduced in 0.4.3, whereby loading
2702
an already-persistent instance mapped with joined table
2703
inheritance would trigger a useless "secondary" load from
2704
its joined table, when using the default "select"
2705
polymorphic_fetch. This was due to attributes being
2706
marked as expired during its first load and not getting
2707
unmarked from the previous "secondary" load. Attributes
2708
are now unexpired based on presence in __dict__ after any
2709
load or commit operation succeeds.
2711
- Deprecated Query methods apply_sum(), apply_max(),
2712
apply_min(), apply_avg(). Better methodologies are
2715
- relation() can accept a callable for its first argument,
2716
which returns the class to be related. This is in place
2717
to assist declarative packages to define relations without
2718
classes yet being in place.
2720
- Added a new "higher level" operator called "of_type()":
2721
used in join() as well as with any() and has(), qualifies
2722
the subclass which will be used in filter criterion, e.g.:
2724
query.filter(Company.employees.of_type(Engineer).
2725
any(Engineer.name=='foo'))
2729
query.join(Company.employees.of_type(Engineer)).
2730
filter(Engineer.name=='foo')
2732
- Preventive code against a potential lost-reference bug in
2735
- Expressions used in filter(), filter_by() and others, when
2736
they make usage of a clause generated from a relation
2737
using the identity of a child object (e.g.,
2738
filter(Parent.child==<somechild>)), evaluate the actual
2739
primary key value of <somechild> at execution time so that
2740
the autoflush step of the Query can complete, thereby
2741
populating the PK value of <somechild> in the case that
2742
<somechild> was pending.
2744
- setting the relation()-level order by to a column in the
2745
many-to-many "secondary" table will now work with eager
2746
loading, previously the "order by" wasn't aliased against
2747
the secondary table's alias.
2749
- Synonyms riding on top of existing descriptors are now
2750
full proxies to those descriptors.
2753
- Invalid SQLite connection URLs now raise an error.
2755
- postgres TIMESTAMP renders correctly [ticket:981]
2757
- postgres PGArray is a "mutable" type by default; when used
2758
with the ORM, mutable-style equality/ copy-on-write
2759
techniques are used to test for changes.
2762
- a new super-small "declarative" extension has been added,
2763
which allows Table and mapper() configuration to take
2764
place inline underneath a class declaration. This
2765
extension differs from ActiveMapper and Elixir in that it
2766
does not redefine any SQLAlchemy semantics at all; literal
2767
Column, Table and relation() constructs are used to define
2768
the class behavior and table definition.
2773
- Added "schema.DDL", an executable free-form DDL statement.
2774
DDLs can be executed in isolation or attached to Table or
2775
MetaData instances and executed automatically when those
2776
objects are created and/or dropped.
2778
- Table columns and constraints can be overridden on a an
2779
existing table (such as a table that was already reflected)
2780
using the 'useexisting=True' flag, which now takes into
2781
account the arguments passed along with it.
2783
- Added a callable-based DDL events interface, adds hooks
2784
before and after Tables and MetaData create and drop.
2786
- Added generative where(<criterion>) method to delete() and
2787
update() constructs which return a new object with criterion
2788
joined to existing criterion via AND, just like
2791
- Added "ilike()" operator to column operations. Compiles to
2792
ILIKE on postgres, lower(x) LIKE lower(y) on all
2793
others. [ticket:727]
2795
- Added "now()" as a generic function; on SQLite, Oracle
2796
and MSSQL compiles as "CURRENT_TIMESTAMP"; "now()" on
2797
all others. [ticket:943]
2799
- The startswith(), endswith(), and contains() operators now
2800
concatenate the wildcard operator with the given operand in
2801
SQL, i.e. "'%' || <bindparam>" in all cases, accept
2802
text('something') operands properly [ticket:962]
2804
- cast() accepts text('something') and other non-literal
2805
operands properly [ticket:962]
2807
- fixed bug in result proxy where anonymously generated
2808
column labels would not be accessible using their straight
2811
- Deferrable constraints can now be defined.
2813
- Added "autocommit=True" keyword argument to select() and
2814
text(), as well as generative autocommit() method on
2815
select(); for statements which modify the database through
2816
some user-defined means other than the usual INSERT/UPDATE/
2817
DELETE etc. This flag will enable "autocommit" behavior
2818
during execution if no transaction is in progress.
2821
- The '.c.' attribute on a selectable now gets an entry for
2822
every column expression in its columns clause. Previously,
2823
"unnamed" columns like functions and CASE statements weren't
2824
getting put there. Now they will, using their full string
2825
representation if no 'name' is available.
2827
- a CompositeSelect, i.e. any union(), union_all(),
2828
intersect(), etc. now asserts that each selectable contains
2829
the same number of columns. This conforms to the
2830
corresponding SQL requirement.
2832
- The anonymous 'label' generated for otherwise unlabeled
2833
functions and expressions now propagates outwards at compile
2834
time for expressions like select([select([func.foo()])]).
2836
- Building on the above ideas, CompositeSelects now build up
2837
their ".c." collection based on the names present in the
2838
first selectable only; corresponding_column() now works
2839
fully for all embedded selectables.
2841
- Oracle and others properly encode SQL used for defaults like
2842
sequences, etc., even if no unicode idents are used since
2843
identifier preparer may return a cached unicode identifier.
2845
- Column and clause comparisons to datetime objects on the
2846
left hand side of the expression now work (d < table.c.col).
2847
(datetimes on the RHS have always worked, the LHS exception
2848
is a quirk of the datetime implementation.)
2851
- Every Session.begin() must now be accompanied by a
2852
corresponding commit() or rollback() unless the session is
2853
closed with Session.close(). This also includes the begin()
2854
which is implicit to a session created with
2855
transactional=True. The biggest change introduced here is
2856
that when a Session created with transactional=True raises
2857
an exception during flush(), you must call
2858
Session.rollback() or Session.close() in order for that
2859
Session to continue after an exception.
2861
- Fixed merge() collection-doubling bug when merging transient
2862
entities with backref'ed collections. [ticket:961]
2864
- merge(dont_load=True) does not accept transient entities,
2865
this is in continuation with the fact that
2866
merge(dont_load=True) does not accept any "dirty" objects
2869
- Added standalone "query" class attribute generated by a
2870
scoped_session. This provides MyClass.query without using
2871
Session.mapper. Use via:
2873
MyClass.query = Session.query_property()
2875
- The proper error message is raised when trying to access
2876
expired instance attributes with no session present
2878
- dynamic_loader() / lazy="dynamic" now accepts and uses
2879
the order_by parameter in the same way in which it works
2882
- Added expire_all() method to Session. Calls expire() for
2883
all persistent instances. This is handy in conjunction
2886
- Instances which have been partially or fully expired will
2887
have their expired attributes populated during a regular
2888
Query operation which affects those objects, preventing a
2889
needless second SQL statement for each instance.
2891
- Dynamic relations, when referenced, create a strong
2892
reference to the parent object so that the query still has a
2893
parent to call against even if the parent is only created
2894
(and otherwise dereferenced) within the scope of a single
2895
expression. [ticket:938]
2897
- Added a mapper() flag "eager_defaults". When set to True,
2898
defaults that are generated during an INSERT or UPDATE
2899
operation are post-fetched immediately, instead of being
2900
deferred until later. This mimics the old 0.3 behavior.
2902
- query.join() can now accept class-mapped attributes as
2903
arguments. These can be used in place or in any combination
2904
with strings. In particular this allows construction of
2905
joins to subclasses on a polymorphic relation, i.e.:
2907
query(Company).join(['employees', Engineer.name])
2909
- query.join() can also accept tuples of attribute name/some
2910
selectable as arguments. This allows construction of joins
2911
*from* subclasses of a polymorphic relation, i.e.:
2915
[('employees', people.join(engineer)), Engineer.name]
2918
- General improvements to the behavior of join() in
2919
conjunction with polymorphic mappers, i.e. joining from/to
2920
polymorphic mappers and properly applying aliases.
2922
- Fixed/improved behavior when a mapper determines the natural
2923
"primary key" of a mapped join, it will more effectively
2924
reduce columns which are equivalent via foreign key
2925
relation. This affects how many arguments need to be sent
2926
to query.get(), among other things. [ticket:933]
2928
- The lazy loader can now handle a join condition where the
2929
"bound" column (i.e. the one that gets the parent id sent as
2930
a bind parameter) appears more than once in the join
2931
condition. Specifically this allows the common task of a
2932
relation() which contains a parent-correlated subquery, such
2933
as "select only the most recent child item". [ticket:946]
2935
- Fixed bug in polymorphic inheritance where an incorrect
2936
exception is raised when base polymorphic_on column does not
2937
correspond to any columns within the local selectable of an
2938
inheriting mapper more than one level deep
2940
- Fixed bug in polymorphic inheritance which made it difficult
2941
to set a working "order_by" on a polymorphic mapper.
2943
- Fixed a rather expensive call in Query that was slowing down
2944
polymorphic queries.
2946
- "Passive defaults" and other "inline" defaults can now be
2947
loaded during a flush() call if needed; in particular, this
2948
allows constructing relations() where a foreign key column
2949
references a server-side-generated, non-primary-key
2950
column. [ticket:954]
2952
- Additional Session transaction fixes/changes:
2953
- Fixed bug with session transaction management: parent
2954
transactions weren't started on the connection when
2955
adding a connection to a nested transaction.
2957
- session.transaction now always refers to the innermost
2958
active transaction, even when commit/rollback are called
2959
directly on the session transaction object.
2961
- Two-phase transactions can now be prepared.
2963
- When preparing a two-phase transaction fails on one
2964
connection, all the connections are rolled back.
2966
- session.close() didn't close all transactions when
2967
nested transactions were used.
2969
- rollback() previously erroneously set the current
2970
transaction directly to the parent of the transaction
2971
that could be rolled back to. Now it rolls back the next
2972
transaction up that can handle it, but sets the current
2973
transaction to it's parent and inactivates the
2974
transactions in between. Inactive transactions can only
2975
be rolled back or closed, any other call results in an
2978
- autoflush for commit() wasn't flushing for simple
2981
- unitofwork flush didn't close the failed transaction
2982
when the session was not in a transaction and commiting
2983
the transaction failed.
2985
- Miscellaneous tickets: [ticket:940] [ticket:964]
2988
- Fixed a variety of hidden and some not-so-hidden
2989
compatibility issues for Python 2.3, thanks to new support
2990
for running the full test suite on 2.3.
2992
- Warnings are now issued as type exceptions.SAWarning.
2995
- Better support for schemas in SQLite (linked in by ATTACH
2996
DATABASE ... AS name). In some cases in the past, schema
2997
names were ommitted from generated SQL for SQLite. This is
3000
- table_names on SQLite now picks up temporary tables as well.
3002
- Auto-detect an unspecified MySQL ANSI_QUOTES mode during
3003
reflection operations, support for changing the mode
3004
midstream. Manual mode setting is still required if no
3007
- Fixed reflection of TIME columns on SQLite.
3009
- Finally added PGMacAddr type to postgres [ticket:580]
3011
- Reflect the sequence associated to a PK field (typically
3012
with a BEFORE INSERT trigger) under Firebird
3014
- Oracle assembles the correct columns in the result set
3015
column mapping when generating a LIMIT/OFFSET subquery,
3016
allows columns to map properly to result sets even if
3017
long-name truncation kicks in [ticket:941]
3019
- MSSQL now includes EXEC in the _is_select regexp, which
3020
should allow row-returning stored procedures to be used.
3022
- MSSQL now includes an experimental implementation of
3023
LIMIT/OFFSET using the ANSI SQL row_number() function, so it
3024
requires MSSQL-2005 or higher. To enable the feature, add
3025
"has_window_funcs" to the keyword arguments for connect, or
3026
add "?has_window_funcs=1" to your dburi query arguments.
3029
- Changed ext.activemapper to use a non-transactional session
3030
for the objectstore.
3032
- Fixed output order of "['a'] + obj.proxied" binary operation
3033
on association-proxied lists.
3038
- sub version numbering scheme changed to suite
3039
setuptools version number rules; easy_install -u
3040
should now get this version over 0.4.2.
3043
- Text type is properly exported now and does not
3044
raise a warning on DDL create; String types with no
3045
length only raise warnings during CREATE TABLE
3048
- new UnicodeText type is added, to specify an
3049
encoded, unlengthed Text type
3051
- fixed bug in union() so that select() statements
3052
which don't derive from FromClause objects can be
3056
- fixed bug with session.dirty when using "mutable
3057
scalars" (such as PickleTypes)
3059
- added a more descriptive error message when flushing
3060
on a relation() that has non-locally-mapped columns
3061
in its primary or secondary join condition
3064
- Fixed reflection of mysql empty string column
3070
- changed name of TEXT to Text since its a "generic"
3071
type; TEXT name is deprecated until 0.5. The
3072
"upgrading" behavior of String to Text when no
3073
length is present is also deprecated until 0.5; will
3074
issue a warning when used for CREATE TABLE
3075
statements (String with no length for SQL expression
3076
purposes is still fine) [ticket:912]
3078
- generative select.order_by(None) / group_by(None)
3079
was not managing to reset order by/group by
3080
criterion, fixed [ticket:924]
3083
- suppressing *all* errors in
3084
InstanceState.__cleanup() now.
3086
- fixed an attribute history bug whereby assigning a
3087
new collection to a collection-based attribute which
3088
already had pending changes would generate incorrect
3089
history [ticket:922]
3091
- fixed delete-orphan cascade bug whereby setting the
3092
same object twice to a scalar attribute could log it
3093
as an orphan [ticket:925]
3095
- Fixed cascades on a += assignment to a list-based
3098
- synonyms can now be created against props that don't
3099
exist yet, which are later added via add_property().
3100
This commonly includes backrefs. (i.e. you can make
3101
synonyms for backrefs without worrying about the
3102
order of operations) [ticket:919]
3104
- fixed bug which could occur with polymorphic "union"
3105
mapper which falls back to "deferred" loading of
3108
- the "columns" collection on a mapper/mapped class
3109
(i.e. 'c') is against the mapped table, not the
3110
select_table in the case of polymorphic "union"
3111
loading (this shouldn't be noticeable).
3114
- '+', '*', '+=' and '*=' support for association
3118
- mssql - narrowed down the test for "date"/"datetime"
3119
in MSDate/ MSDateTime subclasses so that incoming
3120
"datetime" objects don't get mis-interpreted as
3121
"date" objects and vice versa, [ticket:923]
3127
- fixed fairly critical bug whereby the same instance could be listed
3128
more than once in the unitofwork.new collection; most typically
3129
reproduced when using a combination of inheriting mappers and
3130
ScopedSession.mapper, as the multiple __init__ calls per instance
3131
could save() the object with distinct _state objects
3133
- added very rudimentary yielding iterator behavior to Query. Call
3134
query.yield_per(<number of rows>) and evaluate the Query in an
3135
iterative context; every collection of N rows will be packaged up
3136
and yielded. Use this method with extreme caution since it does
3137
not attempt to reconcile eagerly loaded collections across
3138
result batch boundaries, nor will it behave nicely if the same
3139
instance occurs in more than one batch. This means that an eagerly
3140
loaded collection will get cleared out if it's referenced in more than
3141
one batch, and in all cases attributes will be overwritten on instances
3142
that occur in more than one batch.
3144
- Fixed in-place set mutation operators for set collections and association
3145
proxied sets. [ticket:920]
3148
- Fixed the missing call to subtype result processor for the PGArray
3154
- generic functions ! we introduce a database of known SQL functions, such
3155
as current_timestamp, coalesce, and create explicit function objects
3156
representing them. These objects have constrained argument lists, are
3157
type aware, and can compile in a dialect-specific fashion. So saying
3158
func.char_length("foo", "bar") raises an error (too many args),
3159
func.coalesce(datetime.date(2007, 10, 5), datetime.date(2005, 10, 15))
3160
knows that its return type is a Date. We only have a few functions
3161
represented so far but will continue to add to the system [ticket:615]
3163
- auto-reconnect support improved; a Connection can now automatically
3164
reconnect after its underlying connection is invalidated, without
3165
needing to connect() again from the engine. This allows an ORM session
3166
bound to a single Connection to not need a reconnect.
3167
Open transactions on the Connection must be rolled back after an invalidation
3168
of the underlying connection else an error is raised. Also fixed
3169
bug where disconnect detect was not being called for cursor(), rollback(),
3172
- added new flag to String and create_engine(),
3173
assert_unicode=(True|False|'warn'|None). Defaults to `False` or `None` on
3174
create_engine() and String, `'warn'` on the Unicode type. When `True`,
3175
results in all unicode conversion operations raising an exception when a
3176
non-unicode bytestring is passed as a bind parameter. 'warn' results
3177
in a warning. It is strongly advised that all unicode-aware applications
3178
make proper use of Python unicode objects (i.e. u'hello' and not 'hello')
3179
so that data round trips accurately.
3181
- generation of "unique" bind parameters has been simplified to use the same
3182
"unique identifier" mechanisms as everything else. This doesn't affect
3183
user code, except any code that might have been hardcoded against the generated
3184
names. Generated bind params now have the form "<paramname>_<num>",
3185
whereas before only the second bind of the same name would have this form.
3187
- select().as_scalar() will raise an exception if the select does not have
3188
exactly one expression in its columns clause.
3190
- bindparam() objects themselves can be used as keys for execute(), i.e.
3191
statement.execute({bind1:'foo', bind2:'bar'})
3193
- added new methods to TypeDecorator, process_bind_param() and
3194
process_result_value(), which automatically take advantage of the processing
3195
of the underlying type. Ideal for using with Unicode or Pickletype.
3196
TypeDecorator should now be the primary way to augment the behavior of any
3197
existing type including other TypeDecorator subclasses such as PickleType.
3199
- selectables (and others) will issue a warning when two columns in
3200
their exported columns collection conflict based on name.
3202
- tables with schemas can still be used in sqlite, firebird,
3203
schema name just gets dropped [ticket:890]
3205
- changed the various "literal" generation functions to use an anonymous
3206
bind parameter. not much changes here except their labels now look
3207
like ":param_1", ":param_2" instead of ":literal"
3209
- column labels in the form "tablename.columname", i.e. with a dot, are now
3212
- from_obj keyword argument to select() can be a scalar or a list.
3215
- a major behavioral change to collection-based backrefs: they no
3216
longer trigger lazy loads ! "reverse" adds and removes
3217
are queued up and are merged with the collection when it is
3218
actually read from and loaded; but do not trigger a load beforehand.
3219
For users who have noticed this behavior, this should be much more
3220
convenient than using dynamic relations in some cases; for those who
3221
have not, you might notice your apps using a lot fewer queries than
3222
before in some situations. [ticket:871]
3224
- mutable primary key support is added. primary key columns can be
3225
changed freely, and the identity of the instance will change upon
3226
flush. In addition, update cascades of foreign key referents (primary
3227
key or not) along relations are supported, either in tandem with the
3228
database's ON UPDATE CASCADE (required for DB's like Postgres) or
3229
issued directly by the ORM in the form of UPDATE statements, by setting
3230
the flag "passive_cascades=False".
3232
- inheriting mappers now inherit the MapperExtensions of their parent
3233
mapper directly, so that all methods for a particular MapperExtension
3234
are called for subclasses as well. As always, any MapperExtension
3235
can return either EXT_CONTINUE to continue extension processing
3236
or EXT_STOP to stop processing. The order of mapper resolution is:
3237
<extensions declared on the classes mapper> <extensions declared on the
3238
classes' parent mapper> <globally declared extensions>.
3240
Note that if you instantiate the same extension class separately
3241
and then apply it individually for two mappers in the same inheritance
3242
chain, the extension will be applied twice to the inheriting class,
3243
and each method will be called twice.
3245
To apply a mapper extension explicitly to each inheriting class but
3246
have each method called only once per operation, use the same
3247
instance of the extension for both mappers.
3250
- MapperExtension.before_update() and after_update() are now called
3251
symmetrically; previously, an instance that had no modified column
3252
attributes (but had a relation() modification) could be called with
3253
before_update() but not after_update() [ticket:907]
3255
- columns which are missing from a Query's select statement
3256
now get automatically deferred during load.
3258
- mapped classes which extend "object" and do not provide an
3259
__init__() method will now raise TypeError if non-empty *args
3260
or **kwargs are present at instance construction time (and are
3261
not consumed by any extensions such as the scoped_session mapper),
3262
consistent with the behavior of normal Python classes [ticket:908]
3264
- fixed Query bug when filter_by() compares a relation against None
3267
- improved support for pickling of mapped entities. Per-instance
3268
lazy/deferred/expired callables are now serializable so that
3269
they serialize and deserialize with _state.
3271
- new synonym() behavior: an attribute will be placed on the mapped
3272
class, if one does not exist already, in all cases. if a property
3273
already exists on the class, the synonym will decorate the property
3274
with the appropriate comparison operators so that it can be used in in
3275
column expressions just like any other mapped attribute (i.e. usable in
3276
filter(), etc.) the "proxy=True" flag is deprecated and no longer means
3277
anything. Additionally, the flag "map_column=True" will automatically
3278
generate a ColumnProperty corresponding to the name of the synonym,
3279
i.e.: 'somename':synonym('_somename', map_column=True) will map the
3280
column named 'somename' to the attribute '_somename'. See the example
3281
in the mapper docs. [ticket:801]
3283
- Query.select_from() now replaces all existing FROM criterion with
3284
the given argument; the previous behavior of constructing a list
3285
of FROM clauses was generally not useful as is required
3286
filter() calls to create join criterion, and new tables introduced
3287
within filter() already add themselves to the FROM clause. The
3288
new behavior allows not just joins from the main table, but select
3289
statements as well. Filter criterion, order bys, eager load
3290
clauses will be "aliased" against the given statement.
3292
- this month's refactoring of attribute instrumentation changes
3293
the "copy-on-load" behavior we've had since midway through 0.3
3294
with "copy-on-modify" in most cases. This takes a sizable chunk
3295
of latency out of load operations and overall does less work
3296
as only attributes which are actually modified get their
3297
"committed state" copied. Only "mutable scalar" attributes
3298
(i.e. a pickled object or other mutable item), the reason for
3299
the copy-on-load change in the first place, retain the old
3302
- a slight behavioral change to attributes is, del'ing an attribute
3303
does *not* cause the lazyloader of that attribute to fire off again;
3304
the "del" makes the effective value of the attribute "None". To
3305
re-trigger the "loader" for an attribute, use
3306
session.expire(instance, [attrname]).
3308
- query.filter(SomeClass.somechild == None), when comparing
3309
a many-to-one property to None, properly generates "id IS NULL"
3310
including that the NULL is on the right side.
3312
- query.order_by() takes into account aliased joins, i.e.
3313
query.join('orders', aliased=True).order_by(Order.id)
3315
- eagerload(), lazyload(), eagerload_all() take an optional
3316
second class-or-mapper argument, which will select the mapper
3317
to apply the option towards. This can select among other
3318
mappers which were added using add_entity().
3320
- eagerloading will work with mappers added via add_entity().
3322
- added "cascade delete" behavior to "dynamic" relations just like
3323
that of regular relations. if passive_deletes flag (also just added)
3324
is not set, a delete of the parent item will trigger a full load of
3325
the child items so that they can be deleted or updated accordingly.
3327
- also with dynamic, implemented correct count() behavior as well
3328
as other helper methods.
3330
- fix to cascades on polymorphic relations, such that cascades
3331
from an object to a polymorphic collection continue cascading
3332
along the set of attributes specific to each element in the collection.
3334
- query.get() and query.load() do not take existing filter or other
3335
criterion into account; these methods *always* look up the given id
3336
in the database or return the current instance from the identity map,
3337
disregarding any existing filter, join, group_by or other criterion
3338
which has been configured. [ticket:893]
3340
- added support for version_id_col in conjunction with inheriting mappers.
3341
version_id_col is typically set on the base mapper in an inheritance
3342
relationship where it takes effect for all inheriting mappers.
3345
- relaxed rules on column_property() expressions having labels; any
3346
ColumnElement is accepted now, as the compiler auto-labels non-labeled
3347
ColumnElements now. a selectable, like a select() statement, still
3348
requires conversion to ColumnElement via as_scalar() or label().
3350
- fixed backref bug where you could not del instance.attr if attr
3353
- several ORM attributes have been removed or made private:
3354
mapper.get_attr_by_column(), mapper.set_attr_by_column(),
3355
mapper.pks_by_table, mapper.cascade_callable(),
3356
MapperProperty.cascade_callable(), mapper.canload(),
3357
mapper.save_obj(), mapper.delete_obj(), mapper._mapper_registry,
3358
attributes.AttributeManager
3360
- Assigning an incompatible collection type to a relation attribute now
3361
raises TypeError instead of sqlalchemy's ArgumentError.
3363
- Bulk assignment of a MappedCollection now raises an error if a key in the
3364
incoming dictionary does not match the key that the collection's keyfunc
3365
would use for that value. [ticket:886]
3367
- Custom collections can now specify a @converter method to translate
3368
objects used in "bulk" assignment into a stream of values, as in::
3370
obj.col = [newval1, newval2]
3372
obj.dictcol = {'foo': newval1, 'bar': newval2}
3374
The MappedCollection uses this hook to ensure that incoming key/value
3375
pairs are sane from the collection's perspective.
3377
- fixed endless loop issue when using lazy="dynamic" on both
3378
sides of a bi-directional relationship [ticket:872]
3380
- more fixes to the LIMIT/OFFSET aliasing applied with Query + eagerloads,
3381
in this case when mapped against a select statement [ticket:904]
3383
- fix to self-referential eager loading such that if the same mapped
3384
instance appears in two or more distinct sets of columns in the same
3385
result set, its eagerly loaded collection will be populated regardless
3386
of whether or not all of the rows contain a set of "eager" columns for
3387
that collection. this would also show up as a KeyError when fetching
3388
results with join_depth turned on.
3390
- fixed bug where Query would not apply a subquery to the SQL when LIMIT
3391
was used in conjunction with an inheriting mapper where the eager
3392
loader was only in the parent mapper.
3394
- clarified the error message which occurs when you try to update()
3395
an instance with the same identity key as an instance already present
3398
- some clarifications and fixes to merge(instance, dont_load=True).
3399
fixed bug where lazy loaders were getting disabled on returned instances.
3400
Also, we currently do not support merging an instance which has uncommitted
3401
changes on it, in the case that dont_load=True is used....this will
3402
now raise an error. This is due to complexities in merging the
3403
"committed state" of the given instance to correctly correspond to the
3404
newly copied instance, as well as other modified state.
3405
Since the use case for dont_load=True is caching, the given instances
3406
shouldn't have any uncommitted changes on them anyway.
3407
We also copy the instances over without using any events now, so that
3408
the 'dirty' list on the new session remains unaffected.
3410
- fixed bug which could arise when using session.begin_nested() in conjunction
3411
with more than one level deep of enclosing session.begin() statements
3413
- fixed session.refresh() with instance that has custom entity_name
3418
- sqlite SLDate type will not erroneously render "microseconds" portion
3419
of a datetime or time object.
3422
- added disconnect detection support for Oracle
3423
- some cleanup to binary/raw types so that cx_oracle.LOB is detected
3424
on an ad-hoc basis [ticket:902]
3427
- PyODBC no longer has a global "set nocount on".
3428
- Fix non-identity integer PKs on autload [ticket:824]
3429
- Better support for convert_unicode [ticket:839]
3430
- Less strict date conversion for pyodbc/adodbapi [ticket:842]
3431
- Schema-qualified tables / autoload [ticket:901]
3435
- does properly reflect domains (partially fixing [ticket:410]) and
3438
- reverted to use default poolclass (was set to SingletonThreadPool in
3439
0.4.0 [3562] for test purposes)
3441
- map func.length() to 'char_length' (easily overridable with the UDF
3442
'strlen' on old versions of Firebird)
3450
- the "shortname" keyword parameter on bindparam() has been
3453
- Added contains operator (generates a "LIKE %<other>%" clause).
3455
- anonymous column expressions are automatically labeled.
3456
e.g. select([x* 5]) produces "SELECT x * 5 AS anon_1".
3457
This allows the labelname to be present in the cursor.description
3458
which can then be appropriately matched to result-column processing
3459
rules. (we can't reliably use positional tracking for result-column
3460
matches since text() expressions may represent multiple columns).
3462
- operator overloading is now controlled by TypeEngine objects - the
3463
one built-in operator overload so far is String types overloading
3464
'+' to be the string concatenation operator.
3465
User-defined types can also define their own operator overloading
3466
by overriding the adapt_operator(self, op) method.
3468
- untyped bind parameters on the right side of a binary expression
3469
will be assigned the type of the left side of the operation, to better
3470
enable the appropriate bind parameter processing to take effect
3473
- Removed regular expression step from most statement compilations.
3474
Also fixes [ticket:833]
3476
- Fixed empty (zero column) sqlite inserts, allowing inserts on
3477
autoincrementing single column tables.
3479
- Fixed expression translation of text() clauses; this repairs various
3480
ORM scenarios where literal text is used for SQL expressions
3482
- Removed ClauseParameters object; compiled.params returns a regular
3483
dictionary now, as well as result.last_inserted_params() /
3484
last_updated_params().
3486
- Fixed INSERT statements w.r.t. primary key columns that have
3487
SQL-expression based default generators on them; SQL expression
3488
executes inline as normal but will not trigger a "postfetch" condition
3489
for the column, for those DB's who provide it via cursor.lastrowid
3491
- func. objects can be pickled/unpickled [ticket:844]
3493
- rewrote and simplified the system used to "target" columns across
3494
selectable expressions. On the SQL side this is represented by the
3495
"corresponding_column()" method. This method is used heavily by the ORM
3496
to "adapt" elements of an expression to similar, aliased expressions,
3497
as well as to target result set columns originally bound to a
3498
table or selectable to an aliased, "corresponding" expression. The new
3499
rewrite features completely consistent and accurate behavior.
3501
- Added a field ("info") for storing arbitrary data on schema items
3504
- The "properties" collection on Connections has been renamed "info" to
3505
match schema's writable collections. Access is still available via
3506
the "properties" name until 0.5.
3508
- fixed the close() method on Transaction when using strategy='threadlocal'
3510
- fix to compiled bind parameters to not mistakenly populate None
3513
- <Engine|Connection>._execute_clauseelement becomes a public method
3514
Connectable.execute_clauseelement
3517
- eager loading with LIMIT/OFFSET applied no longer adds the primary
3518
table joined to a limited subquery of itself; the eager loads now
3519
join directly to the subquery which also provides the primary table's
3520
columns to the result set. This eliminates a JOIN from all eager loads
3521
with LIMIT/OFFSET. [ticket:843]
3523
- session.refresh() and session.expire() now support an additional argument
3524
"attribute_names", a list of individual attribute keynames to be refreshed
3525
or expired, allowing partial reloads of attributes on an already-loaded
3526
instance. [ticket:802]
3528
- added op() operator to instrumented attributes; i.e.
3529
User.name.op('ilike')('%somename%') [ticket:767]
3531
- Mapped classes may now define __eq__, __hash__, and __nonzero__ methods
3532
with arbitrary semantics. The orm now handles all mapped instances on
3533
an identity-only basis. (e.g. 'is' vs '==') [ticket:676]
3535
- the "properties" accessor on Mapper is removed; it now throws an informative
3536
exception explaining the usage of mapper.get_property() and
3537
mapper.iterate_properties
3539
- added having() method to Query, applies HAVING to the generated statement
3540
in the same way as filter() appends to the WHERE clause.
3542
- The behavior of query.options() is now fully based on paths, i.e. an
3543
option such as eagerload_all('x.y.z.y.x') will apply eagerloading to
3544
only those paths, i.e. and not 'x.y.x'; eagerload('children.children')
3545
applies only to exactly two-levels deep, etc. [ticket:777]
3547
- PickleType will compare using `==` when set up with mutable=False,
3548
and not the `is` operator. To use `is` or any other comparator, send
3549
in a custom comparison function using PickleType(comparator=my_custom_comparator).
3551
- query doesn't throw an error if you use distinct() and an order_by()
3552
containing UnaryExpressions (or other) together [ticket:848]
3554
- order_by() expressions from joined tables are properly added to columns
3555
clause when using distinct() [ticket:786]
3557
- fixed error where Query.add_column() would not accept a class-bound
3558
attribute as an argument; Query also raises an error if an invalid
3559
argument was sent to add_column() (at instances() time) [ticket:858]
3561
- added a little more checking for garbage-collection dereferences in
3562
InstanceState.__cleanup() to reduce "gc ignored" errors on app
3565
- The session API has been solidified:
3567
- It's an error to session.save() an object which is already
3568
persistent [ticket:840]
3570
- It's an error to session.delete() an object which is *not*
3573
- session.update() and session.delete() raise an error when updating
3574
or deleting an instance that is already in the session with a
3577
- The session checks more carefully when determining "object X already
3578
in another session"; e.g. if you pickle a series of objects and
3579
unpickle (i.e. as in a Pylons HTTP session or similar), they can go
3580
into a new session without any conflict
3582
- merge() includes a keyword argument "dont_load=True". setting this
3583
flag will cause the merge operation to not load any data from the
3584
database in response to incoming detached objects, and will accept
3585
the incoming detached object as though it were already present in
3586
that session. Use this to merge detached objects from external
3587
caching systems into the session.
3589
- Deferred column attributes no longer trigger a load operation when the
3590
attribute is assigned to. In those cases, the newly assigned value
3591
will be present in the flushes' UPDATE statement unconditionally.
3593
- Fixed a truncation error when re-assigning a subset of a collection
3594
(obj.relation = obj.relation[1:]) [ticket:834]
3596
- De-cruftified backref configuration code, backrefs which step on
3597
existing properties now raise an error [ticket:832]
3599
- Improved behavior of add_property() etc., fixed [ticket:831] involving
3602
- Fixed clear_mappers() behavior to better clean up after itself.
3604
- Fix to "row switch" behavior, i.e. when an INSERT/DELETE is combined
3605
into a single UPDATE; many-to-many relations on the parent object
3606
update properly. [ticket:841]
3608
- Fixed __hash__ for association proxy- these collections are unhashable,
3609
just like their mutable Python counterparts.
3611
- Added proxying of save_or_update, __contains__ and __iter__ methods for
3614
- fixed very hard-to-reproduce issue where by the FROM clause of Query
3615
could get polluted by certain generative calls [ticket:852]
3619
- Added experimental support for MaxDB (versions >= 7.6.03.007 only).
3621
- oracle will now reflect "DATE" as an OracleDateTime column, not
3624
- added awareness of schema name in oracle table_names() function,
3625
fixes metadata.reflect(schema='someschema') [ticket:847]
3627
- MSSQL anonymous labels for selection of functions made deterministic
3629
- sqlite will reflect "DECIMAL" as a numeric column.
3631
- Made access dao detection more reliable [ticket:828]
3633
- Renamed the Dialect attribute 'preexecute_sequences' to
3634
'preexecute_pk_sequences'. An attribute porxy is in place for
3635
out-of-tree dialects using the old name.
3637
- Added test coverage for unknown type reflection. Fixed sqlite/mysql
3638
handling of type reflection for unknown types.
3640
- Added REAL for mysql dialect (for folks exploiting the
3641
REAL_AS_FLOAT sql mode).
3643
- mysql Float, MSFloat and MSDouble constructed without arguments
3644
now produce no-argument DDL, e.g.'FLOAT'.
3648
- Removed unused util.hash().
3654
- (see 0.4.0beta1 for the start of major changes against 0.3,
3655
as well as http://www.sqlalchemy.org/trac/wiki/WhatsNewIn04 )
3657
- Added initial Sybase support (mxODBC so far) [ticket:785]
3659
- Added partial index support for PostgreSQL. Use the postgres_where keyword
3662
- string-based query param parsing/config file parser understands
3663
wider range of string values for booleans [ticket:817]
3665
- backref remove object operation doesn't fail if the other-side
3666
collection doesn't contain the item, supports noload collections
3669
- removed __len__ from "dynamic" collection as it would require issuing
3670
a SQL "count()" operation, thus forcing all list evaluations to issue
3671
redundant SQL [ticket:818]
3673
- inline optimizations added to locate_dirty() which can greatly speed up
3674
repeated calls to flush(), as occurs with autoflush=True [ticket:816]
3676
- The IdentifierPreprarer's _requires_quotes test is now regex based. Any
3677
out-of-tree dialects that provide custom sets of legal_characters or
3678
illegal_initial_characters will need to move to regexes or override
3681
- Firebird has supports_sane_rowcount and supports_sane_multi_rowcount set
3682
to False due to ticket #370 (right way).
3684
- Improvements and fixes on Firebird reflection:
3685
. FBDialect now mimics OracleDialect, regarding case-sensitivity of TABLE and
3686
COLUMN names (see 'case_sensitive remotion' topic on this current file).
3687
. FBDialect.table_names() doesn't bring system tables (ticket:796).
3688
. FB now reflects Column's nullable property correctly.
3690
- Fixed SQL compiler's awareness of top-level column labels as used
3691
in result-set processing; nested selects which contain the same column
3692
names don't affect the result or conflict with result-column metadata.
3694
- query.get() and related functions (like many-to-one lazyloading)
3695
use compile-time-aliased bind parameter names, to prevent
3696
name conflicts with bind parameters that already exist in the
3699
- Fixed three- and multi-level select and deferred inheritance loading
3700
(i.e. abc inheritance with no select_table), [ticket:795]
3702
- Ident passed to id_chooser in shard.py always a list.
3704
- The no-arg ResultProxy._row_processor() is now the class attribute
3707
- Added support for returning values from inserts and updates for
3708
PostgreSQL 8.2+. [ticket:797]
3710
- PG reflection, upon seeing the default schema name being used explicitly
3711
as the "schema" argument in a Table, will assume that this is the the
3712
user's desired convention, and will explicitly set the "schema" argument
3713
in foreign-key-related reflected tables, thus making them match only
3714
with Table constructors that also use the explicit "schema" argument
3715
(even though its the default schema).
3716
In other words, SA assumes the user is being consistent in this usage.
3718
- fixed sqlite reflection of BOOL/BOOLEAN [ticket:808]
3720
- Added support for UPDATE with LIMIT on mysql.
3722
- null foreign key on a m2o doesn't trigger a lazyload [ticket:803]
3724
- oracle does not implicitly convert to unicode for non-typed result
3725
sets (i.e. when no TypeEngine/String/Unicode type is even being used;
3726
previously it was detecting DBAPI types and converting regardless).
3727
should fix [ticket:800]
3729
- fix to anonymous label generation of long table/column names [ticket:806]
3731
- Firebird dialect now uses SingletonThreadPool as poolclass.
3733
- Firebird now uses dialect.preparer to format sequences names
3735
- Fixed breakage with postgres and multiple two-phase transactions. Two-phase
3736
commits and and rollbacks didn't automatically end up with a new transaction
3737
as the usual dbapi commits/rollbacks do. [ticket:810]
3739
- Added an option to the _ScopedExt mapper extension to not automatically
3740
save new objects to session on object initialization.
3742
- fixed Oracle non-ansi join syntax
3744
- PickleType and Interval types (on db not supporting it natively) are now
3747
- Added Float and Time types to Firebird (FBFloat and FBTime). Fixed
3748
BLOB SUB_TYPE for TEXT and Binary types.
3750
- Changed the API for the in_ operator. in_() now accepts a single argument
3751
that is a sequence of values or a selectable. The old API of passing in
3752
values as varargs still works but is deprecated.
3758
- The Session identity map is now *weak referencing* by default, use
3759
weak_identity_map=False to use a regular dict. The weak dict we are using
3760
is customized to detect instances which are "dirty" and maintain a
3761
temporary strong reference to those instances until changes are flushed.
3763
- Mapper compilation has been reorganized such that most compilation occurs
3764
upon mapper construction. This allows us to have fewer calls to
3765
mapper.compile() and also to allow class-based properties to force a
3766
compilation (i.e. User.addresses == 7 will compile all mappers; this is
3767
[ticket:758]). The only caveat here is that an inheriting mapper now
3768
looks for its inherited mapper upon construction; so mappers within
3769
inheritance relationships need to be constructed in inheritance order
3770
(which should be the normal case anyway).
3772
- added "FETCH" to the keywords detected by Postgres to indicate a
3773
result-row holding statement (i.e. in addition to "SELECT").
3775
- Added full list of SQLite reserved keywords so that they get escaped
3778
- Tightened up the relationship between the Query's generation of "eager
3779
load" aliases, and Query.instances() which actually grabs the eagerly
3780
loaded rows. If the aliases were not specifically generated for that
3781
statement by EagerLoader, the EagerLoader will not take effect when the
3782
rows are fetched. This prevents columns from being grabbed accidentally
3783
as being part of an eager load when they were not meant for such, which
3784
can happen with textual SQL as well as some inheritance situations. It's
3785
particularly important since the "anonymous aliasing" of columns uses
3786
simple integer counts now to generate labels.
3788
- Removed "parameters" argument from clauseelement.compile(), replaced with
3789
"column_keys". The parameters sent to execute() only interact with the
3790
insert/update statement compilation process in terms of the column names
3791
present but not the values for those columns. Produces more consistent
3792
execute/executemany behavior, simplifies things a bit internally.
3794
- Added 'comparator' keyword argument to PickleType. By default, "mutable"
3795
PickleType does a "deep compare" of objects using their dumps()
3796
representation. But this doesn't work for dictionaries. Pickled objects
3797
which provide an adequate __eq__() implementation can be set up with
3798
"PickleType(comparator=operator.eq)" [ticket:560]
3800
- Added session.is_modified(obj) method; performs the same "history"
3801
comparison operation as occurs within a flush operation; setting
3802
include_collections=False gives the same result as is used when the flush
3803
determines whether or not to issue an UPDATE for the instance's row.
3805
- Added "schema" argument to Sequence; use this with Postgres /Oracle when
3806
the sequence is located in an alternate schema. Implements part of
3807
[ticket:584], should fix [ticket:761].
3809
- Fixed reflection of the empty string for mysql enums.
3811
- Changed MySQL dialect to use the older LIMIT <offset>, <limit> syntax
3812
instead of LIMIT <l> OFFSET <o> for folks using 3.23. [ticket:794]
3814
- Added 'passive_deletes="all"' flag to relation(), disables all nulling-out
3815
of foreign key attributes during a flush where the parent object is
3818
- Column defaults and onupdates, executing inline, will add parenthesis for
3819
subqueries and other parenthesis-requiring expressions
3821
- The behavior of String/Unicode types regarding that they auto-convert to
3822
TEXT/CLOB when no length is present now occurs *only* for an exact type of
3823
String or Unicode with no arguments. If you use VARCHAR or NCHAR
3824
(subclasses of String/Unicode) with no length, they will be interpreted by
3825
the dialect as VARCHAR/NCHAR; no "magic" conversion happens there. This
3826
is less surprising behavior and in particular this helps Oracle keep
3827
string-based bind parameters as VARCHARs and not CLOBs [ticket:793].
3829
- Fixes to ShardedSession to work with deferred columns [ticket:771].
3831
- User-defined shard_chooser() function must accept "clause=None" argument;
3832
this is the ClauseElement passed to session.execute(statement) and can be
3833
used to determine correct shard id (since execute() doesn't take an
3836
- Adjusted operator precedence of NOT to match '==' and others, so that
3837
~(x <operator> y) produces NOT (x <op> y), which is better compatible
3838
with older MySQL versions. [ticket:764]. This doesn't apply to "~(x==y)"
3839
as it does in 0.3 since ~(x==y) compiles to "x != y", but still applies
3840
to operators like BETWEEN.
3842
- Other tickets: [ticket:768], [ticket:728], [ticket:779], [ticket:757]
3847
- Connection pool fixes; the better performance of beta4 remains but fixes
3848
"connection overflow" and other bugs which were present (like
3851
- Fixed bugs in determining proper sync clauses from custom inherit
3852
conditions. [ticket:769]
3854
- Extended 'engine_from_config' coercion for QueuePool size / overflow.
3857
- mysql views can be reflected again. [ticket:748]
3859
- AssociationProxy can now take custom getters and setters.
3861
- Fixed malfunctioning BETWEEN in orm queries.
3863
- Fixed OrderedProperties pickling [ticket:762]
3865
- SQL-expression defaults and sequences now execute "inline" for all
3866
non-primary key columns during an INSERT or UPDATE, and for all columns
3867
during an executemany()-style call. inline=True flag on any insert/update
3868
statement also forces the same behavior with a single execute().
3869
result.postfetch_cols() is a collection of columns for which the previous
3870
single insert or update statement contained a SQL-side default expression.
3872
- Fixed PG executemany() behavior, [ticket:759]
3874
- postgres reflects tables with autoincrement=False for primary key columns
3875
which have no defaults.
3877
- postgres no longer wraps executemany() with individual execute() calls,
3878
instead favoring performance. "rowcount"/"concurrency" checks with
3879
deleted items (which use executemany) are disabled with PG since psycopg2
3880
does not report proper rowcount for executemany().
3893
- Tidied up what ends up in your namespace when you 'from sqlalchemy import *':
3895
- 'table' and 'column' are no longer imported. They remain available by
3896
direct reference (as in 'sql.table' and 'sql.column') or a glob import
3897
from the sql package. It was too easy to accidentally use a
3898
sql.expressions.table instead of schema.Table when just starting out
3899
with SQLAlchemy, likewise column.
3901
- Internal-ish classes like ClauseElement, FromClause, NullTypeEngine,
3902
etc., are also no longer imported into your namespace
3904
- The 'Smallinteger' compatiblity name (small i!) is no longer imported,
3905
but remains in schema.py for now. SmallInteger (big I!) is still
3908
- The connection pool uses a "threadlocal" strategy internally to return
3909
the same connection already bound to a thread, for "contextual" connections;
3910
these are the connections used when you do a "connectionless" execution
3911
like insert().execute(). This is like a "partial" version of the
3912
"threadlocal" engine strategy but without the thread-local transaction part
3913
of it. We're hoping it reduces connection pool overhead as well as
3914
database usage. However, if it proves to impact stability in a negative way,
3915
we'll roll it right back.
3917
- Fix to bind param processing such that "False" values (like blank strings)
3918
still get processed/encoded.
3920
- Fix to select() "generative" behavior, such that calling column(),
3921
select_from(), correlate(), and with_prefix() does not modify the
3922
original select object [ticket:752]
3924
- Added a "legacy" adapter to types, such that user-defined TypeEngine
3925
and TypeDecorator classes which define convert_bind_param() and/or
3926
convert_result_value() will continue to function. Also supports
3927
calling the super() version of those methods.
3929
- Added session.prune(), trims away instances cached in a session that
3930
are no longer referenced elsewhere. (A utility for strong-ref
3933
- Added close() method to Transaction. Closes out a transaction using
3934
rollback if it's the outermost transaction, otherwise just ends
3935
without affecting the outer transaction.
3937
- Transactional and non-transactional Session integrates better with
3938
bound connection; a close() will ensure that connection
3939
transactional state is the same as that which existed on it before
3940
being bound to the Session.
3942
- Modified SQL operator functions to be module-level operators,
3943
allowing SQL expressions to be pickleable. [ticket:735]
3945
- Small adjustment to mapper class.__init__ to allow for Py2.6
3946
object.__init__() behavior.
3948
- Fixed 'prefix' argument for select()
3950
- Connection.begin() no longer accepts nested=True, this logic is now
3951
all in begin_nested().
3953
- Fixes to new "dynamic" relation loader involving cascades
3963
- SQL types optimization:
3965
- New performance tests show a combined mass-insert/mass-select test as
3966
having 68% fewer function calls than the same test run against 0.3.
3968
- General performance improvement of result set iteration is around 10-20%.
3970
- In types.AbstractType, convert_bind_param() and convert_result_value()
3971
have migrated to callable-returning bind_processor() and
3972
result_processor() methods. If no callable is returned, no pre/post
3973
processing function is called.
3975
- Hooks added throughout base/sql/defaults to optimize the calling of bind
3976
aram/result processors so that method call overhead is minimized.
3978
- Support added for executemany() scenarios such that unneeded "last row id"
3979
logic doesn't kick in, parameters aren't excessively traversed.
3981
- Added 'inherit_foreign_keys' arg to mapper().
3983
- Added support for string date passthrough in sqlite.
3995
- mssql improvements.
3997
- oracle improvements.
3999
- Auto-commit after LOAD DATA INFILE for mysql.
4001
- A rudimental SessionExtension class has been added, allowing user-defined
4002
functionality to take place at flush(), commit(), and rollback() boundaries.
4004
- Added engine_from_config() function for helping to create_engine() from an
4007
- base_mapper() becomes a plain attribute.
4009
- session.execute() and scalar() can search for a Table with which to bind from
4010
using the given ClauseElement.
4012
- Session automatically extrapolates tables from mappers with binds, also uses
4013
base_mapper so that inheritance hierarchies bind automatically.
4015
- Moved ClauseVisitor traversal back to inlined non-recursive.
4029
- Speed! Along with recent speedups to ResultProxy, total number of function
4030
calls significantly reduced for large loads.
4032
- test/perf/masseagerload.py reports 0.4 as having the fewest number of
4033
function calls across all SA versions (0.1, 0.2, and 0.3).
4035
- New collection_class api and implementation [ticket:213]. Collections are
4036
now instrumented via decorations rather than proxying. You can now have
4037
collections that manage their own membership, and your class instance will
4038
be directly exposed on the relation property. The changes are transparent
4041
- InstrumentedList (as it was) is removed, and relation properties no
4042
longer have 'clear()', '.data', or any other added methods beyond those
4043
provided by the collection type. You are free, of course, to add them to
4046
- __setitem__-like assignments now fire remove events for the existing
4049
- dict-likes used as collection classes no longer need to change __iter__
4050
semantics- itervalues() is used by default instead. This is a backwards
4051
incompatible change.
4053
- Subclassing dict for a mapped collection is no longer needed in most
4054
cases. orm.collections provides canned implementations that key objects
4055
by a specified column or a custom function of your choice.
4057
- Collection assignment now requires a compatible type- assigning None to
4058
clear a collection or assigning a list to a dict collection will now
4059
raise an argument error.
4061
- AttributeExtension moved to interfaces, and .delete is now .remove The
4062
event method signature has also been swapped around.
4064
- Major overhaul for Query:
4066
- All selectXXX methods are deprecated. Generative methods are now the
4067
standard way to do things, i.e. filter(), filter_by(), all(), one(),
4068
etc. Deprecated methods are docstring'ed with their new replacements.
4070
- Class-level properties are now usable as query elements... no more
4071
'.c.'! "Class.c.propname" is now superceded by "Class.propname". All
4072
clause operators are supported, as well as higher level operators such
4073
as Class.prop==<some instance> for scalar attributes,
4074
Class.prop.contains(<some instance>) and Class.prop.any(<some
4075
expression>) for collection-based attributes (all are also
4076
negatable). Table-based column expressions as well as columns mounted
4077
on mapped classes via 'c' are of course still fully available and can be
4078
freely mixed with the new attributes. [ticket:643]
4080
- Removed ancient query.select_by_attributename() capability.
4082
- The aliasing logic used by eager loading has been generalized, so that
4083
it also adds full automatic aliasing support to Query. It's no longer
4084
necessary to create an explicit Alias to join to the same tables
4085
multiple times; *even for self-referential relationships*.
4087
- join() and outerjoin() take arguments "aliased=True". Yhis causes
4088
their joins to be built on aliased tables; subsequent calls to
4089
filter() and filter_by() will translate all table expressions (yes,
4090
real expressions using the original mapped Table) to be that of the
4091
Alias for the duration of that join() (i.e. until reset_joinpoint() or
4092
another join() is called).
4094
- join() and outerjoin() take arguments "id=<somestring>". When used
4095
with "aliased=True", the id can be referenced by add_entity(cls,
4096
id=<somestring>) so that you can select the joined instances even if
4097
they're from an alias.
4099
- join() and outerjoin() now work with self-referential relationships!
4100
Using "aliased=True", you can join as many levels deep as desired,
4101
i.e. query.join(['children', 'children'], aliased=True); filter
4102
criterion will be against the rightmost joined table
4104
- Added query.populate_existing(), marks the query to reload all
4105
attributes and collections of all instances touched in the query,
4106
including eagerly-loaded entities. [ticket:660]
4108
- Added eagerload_all(), allows eagerload_all('x.y.z') to specify eager
4109
loading of all properties in the given path.
4111
- Major overhaul for Session:
4113
- New function which "configures" a session called "sessionmaker()". Send
4114
various keyword arguments to this function once, returns a new class
4115
which creates a Session against that stereotype.
4117
- SessionTransaction removed from "public" API. You now can call begin()/
4118
commit()/rollback() on the Session itself.
4120
- Session also supports SAVEPOINT transactions; call begin_nested().
4122
- Session supports two-phase commit behavior when vertically or
4123
horizontally partitioning (i.e., using more than one engine). Use
4126
- Session flag "transactional=True" produces a session which always places
4127
itself into a transaction when first used. Upon commit(), rollback() or
4128
close(), the transaction ends; but begins again on the next usage.
4130
- Session supports "autoflush=True". This issues a flush() before each
4131
query. Use in conjunction with transactional, and you can just
4132
save()/update() and then query, the new objects will be there. Use
4133
commit() at the end (or flush() if non-transactional) to flush remaining
4136
- New scoped_session() function replaces SessionContext and assignmapper.
4137
Builds onto "sessionmaker()" concept to produce a class whos Session()
4138
construction returns the thread-local session. Or, call all Session
4139
methods as class methods, i.e. Session.save(foo); Session.commit().
4140
just like the old "objectstore" days.
4142
- Added new "binds" argument to Session to support configuration of
4143
multiple binds with sessionmaker() function.
4145
- A rudimental SessionExtension class has been added, allowing
4146
user-defined functionality to take place at flush(), commit(), and
4147
rollback() boundaries.
4149
- Query-based relation()s available with dynamic_loader(). This is a
4150
*writable* collection (supporting append() and remove()) which is also a
4151
live Query object when accessed for reads. Ideal for dealing with very
4152
large collections where only partial loading is desired.
4154
- flush()-embedded inline INSERT/UPDATE expressions. Assign any SQL
4155
expression, like "sometable.c.column + 1", to an instance's attribute.
4156
Upon flush(), the mapper detects the expression and embeds it directly in
4157
the INSERT or UPDATE statement; the attribute gets deferred on the
4158
instance so it loads the new value the next time you access it.
4160
- A rudimental sharding (horizontal scaling) system is introduced. This
4161
system uses a modified Session which can distribute read and write
4162
operations among multiple databases, based on user-defined functions
4163
defining the "sharding strategy". Instances and their dependents can be
4164
distributed and queried among multiple databases based on attribute
4165
values, round-robin approaches or any other user-defined
4166
system. [ticket:618]
4168
- Eager loading has been enhanced to allow even more joins in more places.
4169
It now functions at any arbitrary depth along self-referential and
4170
cyclical structures. When loading cyclical structures, specify
4171
"join_depth" on relation() indicating how many times you'd like the table
4172
to join to itself; each level gets a distinct table alias. The alias
4173
names themselves are generated at compile time using a simple counting
4174
scheme now and are a lot easier on the eyes, as well as of course
4175
completely deterministic. [ticket:659]
4177
- Added composite column properties. This allows you to create a type which
4178
is represented by more than one column, when using the ORM. Objects of
4179
the new type are fully functional in query expressions, comparisons,
4180
query.get() clauses, etc. and act as though they are regular single-column
4181
scalars... except they're not! Use the function composite(cls, *columns)
4182
inside of the mapper's "properties" dict, and instances of cls will be
4183
created/mapped to a single attribute, comprised of the values correponding
4184
to *columns. [ticket:211]
4186
- Improved support for custom column_property() attributes which feature
4187
correlated subqueries, works better with eager loading now.
4189
- Primary key "collapse" behavior; the mapper will analyze all columns in
4190
its given selectable for primary key "equivalence", that is, columns which
4191
are equivalent via foreign key relationship or via an explicit
4192
inherit_condition. primarily for joined-table inheritance scenarios where
4193
different named PK columns in inheriting tables should "collapse" into a
4194
single-valued (or fewer-valued) primary key. Fixes things like
4197
- Joined-table inheritance will now generate the primary key columns of all
4198
inherited classes against the root table of the join only. This implies
4199
that each row in the root table is distinct to a single instance. If for
4200
some rare reason this is not desireable, explicit primary_key settings on
4201
individual mappers will override it.
4203
- When "polymorphic" flags are used with joined-table or single-table
4204
inheritance, all identity keys are generated against the root class of the
4205
inheritance hierarchy; this allows query.get() to work polymorphically
4206
using the same caching semantics as a non-polymorphic get. Note that this
4207
currently does not work with concrete inheritance.
4209
- Secondary inheritance loading: polymorphic mappers can be constructed
4210
*without* a select_table argument. inheriting mappers whose tables were
4211
not represented in the initial load will issue a second SQL query
4212
immediately, once per instance (i.e. not very efficient for large lists),
4213
in order to load the remaining columns.
4215
- Secondary inheritance loading can also move its second query into a
4216
column-level "deferred" load, via the "polymorphic_fetch" argument, which
4217
can be set to 'select' or 'deferred'
4219
- It's now possible to map only a subset of available selectable columns
4220
onto mapper properties, using include_columns/exclude_columns.
4223
- Added undefer_group() MapperOption, sets a set of "deferred" columns
4224
joined by a "group" to load as "undeferred".
4226
- Rewrite of the "deterministic alias name" logic to be part of the SQL
4227
layer, produces much simpler alias and label names more in the style of
4232
- Speed! Clause compilation as well as the mechanics of SQL constructs have
4233
been streamlined and simplified to a signficant degree, for a 20-30%
4234
improvement of the statement construction/compilation overhead of 0.3.
4236
- All "type" keyword arguments, such as those to bindparam(), column(),
4237
Column(), and func.<something>(), renamed to "type_". Those objects still
4238
name their "type" attribute as "type".
4240
- case_sensitive=(True|False) setting removed from schema items, since
4241
checking this state added a lot of method call overhead and there was no
4242
decent reason to ever set it to False. Table and column names which are
4243
all lower case will be treated as case-insenstive (yes we adjust for
4244
Oracle's UPPERCASE style too).
4248
- Added context manager (with statement) support for transactions.
4249
- Added support for two phase commit, works with mysql and postgres so far.
4250
- Added a subtransaction implementation that uses savepoints.
4251
- Added support for savepoints.
4255
- Tables can be reflected from the database en-masse without declaring
4256
them in advance. MetaData(engine, reflect=True) will load all tables
4257
present in the database, or use metadata.reflect() for finer control.
4258
- DynamicMetaData has been renamed to ThreadLocalMetaData
4259
- The ThreadLocalMetaData constructor now takes no arguments.
4260
- BoundMetaData has been removed- regular MetaData is equivalent
4262
- Numeric and Float types now have an "asdecimal" flag; defaults to True for
4263
Numeric, False for Float. When True, values are returned as
4264
decimal.Decimal objects; when False, values are returned as float(). The
4265
defaults of True/False are already the behavior for PG and MySQL's DBAPI
4266
modules. [ticket:646]
4268
- New SQL operator implementation which removes all hardcoded operators from
4269
expression structures and moves them into compilation; allows greater
4270
flexibility of operator compilation; for example, "+" compiles to "||"
4271
when used in a string context, or "concat(a,b)" on MySQL; whereas in a
4272
numeric context it compiles to "+". Fixes [ticket:475].
4274
- "Anonymous" alias and label names are now generated at SQL compilation
4275
time in a completely deterministic fashion... no more random hex IDs
4277
- Significant architectural overhaul to SQL elements (ClauseElement). All
4278
elements share a common "mutability" framework which allows a consistent
4279
approach to in-place modifications of elements as well as generative
4280
behavior. Improves stability of the ORM which makes heavy usage of
4281
mutations to SQL expressions.
4283
- select() and union()'s now have "generative" behavior. Methods like
4284
order_by() and group_by() return a *new* instance - the original instance
4285
is left unchanged. Non-generative methods remain as well.
4287
- The internals of select/union vastly simplified- all decision making
4288
regarding "is subquery" and "correlation" pushed to SQL generation phase.
4289
select() elements are now *never* mutated by their enclosing containers or
4290
by any dialect's compilation process [ticket:52] [ticket:569]
4292
- select(scalar=True) argument is deprecated; use select(..).as_scalar().
4293
The resulting object obeys the full "column" interface and plays better
4296
- Added select().with_prefix('foo') allowing any set of keywords to be
4297
placed before the columns clause of the SELECT [ticket:504]
4299
- Added array slice support to row[<index>] [ticket:686]
4301
- Result sets make a better attempt at matching the DBAPI types present in
4302
cursor.description to the TypeEngine objects defined by the dialect, which
4303
are then used for result-processing. Note this only takes effect for
4304
textual SQL; constructed SQL statements always have an explicit type map.
4306
- Result sets from CRUD operations close their underlying cursor immediately
4307
and will also autoclose the connection if defined for the operation; this
4308
allows more efficient usage of connections for successive CRUD operations
4309
with less chance of "dangling connections".
4311
- Column defaults and onupdate Python functions (i.e. passed to
4312
ColumnDefault) may take zero or one arguments; the one argument is the
4313
ExecutionContext, from which you can call "context.parameters[someparam]"
4314
to access the other bind parameter values affixed to the statement
4315
[ticket:559]. The connection used for the execution is available as well
4316
so that you can pre-execute statements.
4318
- Added "explcit" create/drop/execute support for sequences (i.e. you can
4319
pass a "connectable" to each of those methods on Sequence).
4321
- Better quoting of identifiers when manipulating schemas.
4323
- Standardized the behavior for table reflection where types can't be
4324
located; NullType is substituted instead, warning is raised.
4326
- ColumnCollection (i.e. the 'c' attribute on tables) follows dictionary
4327
semantics for "__contains__" [ticket:606]
4331
- Speed! The mechanics of result processing and bind parameter processing
4332
have been overhauled, streamlined and optimized to issue as little method
4333
calls as possible. Bench tests for mass INSERT and mass rowset iteration
4334
both show 0.4 to be over twice as fast as 0.3, using 68% fewer function
4337
- You can now hook into the pool lifecycle and run SQL statements or other
4338
logic at new each DBAPI connection, pool check-out and check-in.
4340
- Connections gain a .properties collection, with contents scoped to the
4341
lifetime of the underlying DBAPI connection
4343
- Removed auto_close_cursors and disallow_open_cursors arguments from Pool;
4344
reduces overhead as cursors are normally closed by ResultProxy and
4349
- proxyengine is temporarily removed, pending an actually working
4352
- SelectResults has been replaced by Query. SelectResults /
4353
SelectResultsExt still exist but just return a slightly modified Query
4354
object for backwards-compatibility. join_to() method from SelectResults
4355
isn't present anymore, need to use join().
4359
- Table and column names loaded via reflection are now Unicode.
4361
- All standard column types are now supported, including SET.
4363
- Table reflection can now be performed in as little as one round-trip.
4365
- ANSI and ANSI_QUOTES sql modes are now supported.
4367
- Indexes are now reflected.
4371
- Added PGArray datatype for using postgres array datatypes.
4375
- Very rudimental support for OUT parameters added; use sql.outparam(name,
4376
type) to set up an OUT parameter, just like bindparam(); after execution,
4377
values are avaiable via result.out_parameters dictionary. [ticket:507]
4384
- tweak DISTINCT precedence for clauses like
4385
`func.count(t.c.col.distinct())`
4387
- Fixed detection of internal '$' characters in :bind$params [ticket:719]
4389
- [ticket:768] dont assume join criterion consists only of column objects
4391
- adjusted operator precedence of NOT to match '==' and others, so that
4392
~(x==y) produces NOT (x=y), which is compatible with MySQL < 5.0
4393
(doesn't like "NOT x=y") [ticket:764]
4397
- added a check for joining from A->B using join(), along two
4398
different m2m tables. this raises an error in 0.3 but is
4399
possible in 0.4 when aliases are used. [ticket:687]
4401
- fixed small exception throw bug in Session.merge()
4403
- fixed bug where mapper, being linked to a join where one table had
4404
no PK columns, would not detect that the joined table had no PK.
4406
- fixed bugs in determining proper sync clauses from custom inherit
4407
conditions [ticket:769]
4409
- backref remove object operation doesn't fail if the other-side
4410
collection doesn't contain the item, supports noload collections
4415
- fixed another occasional race condition which could occur
4416
when using pool with threadlocal setting
4419
- fixed specification of YEAR columns when generating schema
4423
- added support for TIME columns (simulated using DATETIME) [ticket:679]
4425
- added support for BIGINT, MONEY, SMALLMONEY, UNIQUEIDENTIFIER and
4426
SQL_VARIANT [ticket:721]
4428
- index names are now quoted when dropping from reflected tables
4431
- can now specify a DSN for PyODBC, using a URI like mssql:///?dsn=bob
4435
- when reflecting tables from alternate schemas, the "default" placed upon
4436
the primary key, i.e. usually a sequence name, has the "schema" name
4437
unconditionally quoted, so that schema names which need quoting are fine.
4438
its slightly unnecessary for schema names which don't need quoting
4442
- passthrough for stringified dates
4445
- supports_sane_rowcount() set to False due to ticket #370 (right way).
4446
- fixed reflection of Column's nullable property.
4449
- removed LONG_STRING, LONG_BINARY from "binary" types, so type objects
4450
don't try to read their values as LOB [ticket:622], [ticket:751]
4454
- a new mutex that was added in 0.3.9 causes the pool_timeout
4455
feature to fail during a race condition; threads would
4456
raise TimeoutError immediately with no delay if many threads
4457
push the pool into overflow at the same time. this issue has been
4460
- got connection-bound metadata to work with implicit execution
4461
- foreign key specs can have any chararcter in their identifiers
4463
- added commutativity-awareness to binary clause comparisons to
4464
each other, improves ORM lazy load optimization [ticket:664]
4466
- cleanup to connection-bound sessions, SessionTransaction
4468
- fixed max identifier length (63) [ticket:571]
4473
- better error message for NoSuchColumnError [ticket:607]
4474
- finally figured out how to get setuptools version in, available
4475
as sqlalchemy.__version__ [ticket:428]
4476
- the various "engine" arguments, such as "engine", "connectable",
4477
"engine_or_url", "bind_to", etc. are all present, but deprecated.
4478
they all get replaced by the single term "bind". you also
4479
set the "bind" of MetaData using
4480
metadata.bind = <engine or connection>
4482
- iteration over dict association proxies is now dict-like, not
4483
InstrumentedList-like (e.g. over keys instead of values)
4484
- association proxies no longer bind tightly to source collections
4485
[ticket:597], and are constructed with a thunk instead
4486
- added selectone_by() to assignmapper
4488
- forwards-compatibility with 0.4: added one(), first(), and
4489
all() to Query. almost all Query functionality from 0.4 is
4490
present in 0.3.9 for forwards-compat purposes.
4491
- reset_joinpoint() really really works this time, promise ! lets
4492
you re-join from the root:
4493
query.join(['a', 'b']).filter(<crit>).reset_joinpoint().\
4494
join(['a', 'c']).filter(<some other crit>).all()
4495
in 0.4 all join() calls start from the "root"
4496
- added synchronization to the mapper() construction step, to avoid
4497
thread collisions when pre-existing mappers are compiling in a
4498
different thread [ticket:613]
4499
- a warning is issued by Mapper when two primary key columns of the
4500
same name are munged into a single attribute. this happens frequently
4501
when mapping to joins (or inheritance).
4502
- synonym() properties are fully supported by all Query joining/
4503
with_parent operations [ticket:598]
4504
- fixed very stupid bug when deleting items with many-to-many
4505
uselist=False relations
4506
- remember all that stuff about polymorphic_union ? for
4507
joined table inheritance ? Funny thing...
4508
You sort of don't need it for joined table inheritance, you
4509
can just string all the tables together via outerjoin().
4510
The UNION still applies if concrete tables are involved,
4511
though (since nothing to join them on).
4512
- small fix to eager loading to better work with eager loads
4513
to polymorphic mappers that are using a straight "outerjoin"
4516
- ForeignKey to a table in a schema thats not the default schema
4517
requires the schema to be explicit; i.e. ForeignKey('alt_schema.users.id')
4518
- MetaData can now be constructed with an engine or url as the first
4519
argument, just like BoundMetaData
4520
- BoundMetaData is now deprecated, and MetaData is a direct substitute.
4521
- DynamicMetaData has been renamed to ThreadLocalMetaData. the
4522
DynamicMetaData name is deprecated and is an alias for ThreadLocalMetaData
4523
or a regular MetaData if threadlocal=False
4524
- composite primary key is represented as a non-keyed set to allow for
4525
composite keys consisting of cols with the same name; occurs within a
4526
Join. helps inheritance scenarios formulate correct PK.
4527
- improved ability to get the "correct" and most minimal set of primary key
4528
columns from a join, equating foreign keys and otherwise equated columns.
4529
this is also mostly to help inheritance scenarios formulate the best
4530
choice of primary key columns. [ticket:185]
4531
- added 'bind' argument to Sequence.create()/drop(), ColumnDefault.execute()
4532
- columns can be overridden in a reflected table with a "key"
4533
attribute different than the column's name, including for primary key
4534
columns [ticket:650]
4535
- fixed "ambiguous column" result detection, when dupe col names exist
4536
in a result [ticket:657]
4537
- some enhancements to "column targeting", the ability to match a column
4538
to a "corresponding" column in another selectable. this affects mostly
4539
ORM ability to map to complex joins
4540
- MetaData and all SchemaItems are safe to use with pickle. slow
4541
table reflections can be dumped into a pickled file to be reused later.
4542
Just reconnect the engine to the metadata after unpickling. [ticket:619]
4543
- added a mutex to QueuePool's "overflow" calculation to prevent a race
4544
condition that can bypass max_overflow
4545
- fixed grouping of compound selects to give correct results. will break
4546
on sqlite in some cases, but those cases were producing incorrect
4547
results anyway, sqlite doesn't support grouped compound selects
4549
- fixed precedence of operators so that parenthesis are correctly applied
4551
- calling <column>.in_() (i.e. with no arguments) will return
4552
"CASE WHEN (<column> IS NULL) THEN NULL ELSE 0 END = 1)", so that
4553
NULL or False is returned in all cases, rather than throwing an error
4555
- fixed "where"/"from" criterion of select() to accept a unicode string
4556
in addition to regular string - both convert to text()
4557
- added standalone distinct() function in addition to column.distinct()
4559
- result.last_inserted_ids() should return a list that is identically
4560
sized to the primary key constraint of the table. values that were
4561
"passively" created and not available via cursor.lastrowid will be None.
4562
- long-identifier detection fixed to use > rather than >= for
4563
max ident length [ticket:589]
4564
- fixed bug where selectable.corresponding_column(selectable.c.col)
4565
would not return selectable.c.col, if the selectable is a join
4566
of a table and another join involving the same table. messed
4567
up ORM decision making [ticket:593]
4568
- added Interval type to types.py [ticket:595]
4570
- fixed catching of some errors that imply a dropped connection [ticket:625]
4571
- fixed escaping of the modulo operator [ticket:624]
4572
- added 'fields' to reserved words [ticket:590]
4573
- various reflection enhancement/fixes
4575
- datetime fixes: got subsecond TIMESTAMP to work [ticket:604],
4576
added OracleDate which supports types.Date with only year/month/day
4577
- added dialect flag "auto_convert_lobs", defaults to True; will cause any
4578
LOB objects detected in a result set to be forced into OracleBinary
4579
so that the LOB is read() automatically, if no typemap was present
4580
(i.e., if a textual execute() was issued).
4581
- mod operator '%' produces MOD [ticket:624]
4582
- converts cx_oracle datetime objects to Python datetime.datetime when
4583
Python 2.3 used [ticket:542]
4584
- fixed unicode conversion in Oracle TEXT type
4586
- fixed escaping of the modulo operator [ticket:624]
4587
- added support for reflection of domains [ticket:570]
4588
- types which are missing during reflection resolve to Null type
4589
instead of raising an error
4590
- the fix in "schema" above fixes reflection of foreign keys from an
4591
alt-schema table to a public schema table
4593
- rearranged dialect initialization so it has time to warn about pysqlite1
4595
- sqlite better handles datetime/date/time objects mixed and matched
4596
with various Date/Time/DateTime columns
4597
- string PK column inserts dont get overwritten with OID [ticket:603]
4599
- fix port option handling for pyodbc [ticket:634]
4600
- now able to reflect start and increment values for identity columns
4601
- preliminary support for using scope_identity() with pyodbc
4605
- added detach() to Connection, allows underlying DBAPI connection
4606
to be detached from its pool, closing on dereference/close()
4607
instead of being reused by the pool.
4608
- added invalidate() to Connection, immediately invalidates the
4609
Connection and its underlying DBAPI connection.
4611
- _Label class overrides compare_self to return its ultimate
4612
object. meaning, if you say someexpr.label('foo') == 5, it
4613
produces the correct "someexpr == 5".
4614
- _Label propagates "_hide_froms()" so that scalar selects
4615
behave more properly with regards to FROM clause #574
4616
- fix to long name generation when using oid_column as an order by
4617
(oids used heavily in mapper queries)
4618
- significant speed improvement to ResultProxy, pre-caches
4619
TypeEngine dialect implementations and saves on function calls
4621
- parenthesis are applied to clauses via a new _Grouping
4622
construct. uses operator precedence to more intelligently apply
4623
parenthesis to clauses, provides cleaner nesting of clauses
4624
(doesnt mutate clauses placed in other clauses, i.e. no 'parens'
4626
- added 'modifier' keyword, works like func.<foo> except does not
4627
add parenthesis. e.g. select([modifier.DISTINCT(...)]) etc.
4628
- removed "no group by's in a select thats part of a UNION"
4629
restriction [ticket:578]
4631
- added reset_joinpoint() method to Query, moves the "join point"
4632
back to the starting mapper. 0.4 will change the behavior of
4633
join() to reset the "join point" in all cases so this is an
4634
interim method. for forwards compatibility, ensure joins across
4635
multiple relations are specified using a single join(), i.e.
4636
join(['a', 'b', 'c']).
4637
- fixed bug in query.instances() that wouldnt handle more than
4638
on additional mapper or one additional column.
4639
- "delete-orphan" no longer implies "delete". ongoing effort to
4640
separate the behavior of these two operations.
4641
- many-to-many relationships properly set the type of bind params
4642
for delete operations on the association table
4643
- many-to-many relationships check that the number of rows deleted
4644
from the association table by a delete operation matches the
4646
- session.get() and session.load() propagate **kwargs through to
4648
- fix to polymorphic query which allows the original
4649
polymorphic_union to be embedded into a correlated subquery
4651
- fix to select_by(<propname>=<object instance>) -style joins in
4652
conjunction with many-to-many relationships, bug introduced in
4654
- the "primary_key" argument to mapper() is propagated to the
4655
"polymorphic" mapper. primary key columns in this list get
4656
normalized to that of the mapper's local table.
4657
- restored logging of "lazy loading clause" under
4658
sa.orm.strategies logger, got removed in 0.3.7
4659
- improved support for eagerloading of properties off of mappers
4660
that are mapped to select() statements; i.e. eagerloader is
4661
better at locating the correct selectable with which to attach
4662
its LEFT OUTER JOIN.
4664
- Nearly all MySQL column types are now supported for declaration
4665
and reflection. Added NCHAR, NVARCHAR, VARBINARY, TINYBLOB,
4667
- The sqltypes.Binary passthrough now always builds a BLOB,
4668
avoiding problems with very old database versions
4669
- support for column-level CHARACTER SET and COLLATE declarations,
4670
as well as ASCII, UNICODE, NATIONAL and BINARY shorthand.
4672
- set max identifier length to 31
4673
- supports_sane_rowcount() set to False due to ticket #370.
4674
versioned_id_col feature wont work in FB.
4675
- some execution fixes
4677
- new association proxy implementation, implementing complete
4678
proxies to list, dict and set-based relation collections
4679
- added orderinglist, a custom list class that synchronizes an
4680
object attribute with that object's position in the list
4681
- small fix to SelectResultsExt to not bypass itself during
4683
- added filter(), filter_by() to assignmapper
4687
- warnings module used for issuing warnings (instead of logging)
4688
- cleanup of DBAPI import strategies across all engines
4690
- refactoring of engine internals which reduces complexity,
4691
number of codepaths; places more state inside of ExecutionContext
4692
to allow more dialect control of cursor handling, result sets.
4693
ResultProxy totally refactored and also has two versions of
4694
"buffered" result sets used for different purposes.
4695
- server side cursor support fully functional in postgres
4697
- improved framework for auto-invalidation of connections that have
4698
lost their underlying database, via dialect-specific detection
4699
of exceptions corresponding to that database's disconnect
4700
related error messages. Additionally, when a "connection no
4701
longer open" condition is detected, the entire connection pool
4702
is discarded and replaced with a new instance. #516
4703
- the dialects within sqlalchemy.databases become a setuptools
4704
entry points. loading the built-in database dialects works the
4705
same as always, but if none found will fall back to trying
4706
pkg_resources to load an external module [ticket:521]
4707
- Engine contains a "url" attribute referencing the url.URL object
4708
used by create_engine().
4710
- keys() of result set columns are not lowercased, come back
4711
exactly as they're expressed in cursor.description. note this
4712
causes colnames to be all caps in oracle.
4713
- preliminary support for unicode table names, column names and
4714
SQL statements added, for databases which can support them.
4715
Works with sqlite and postgres so far. Mysql *mostly* works
4716
except the has_table() function does not work. Reflection
4718
- the Unicode type is now a direct subclass of String, which now
4719
contains all the "convert_unicode" logic. This helps the variety
4720
of unicode situations that occur in db's such as MS-SQL to be
4721
better handled and allows subclassing of the Unicode datatype.
4723
- ClauseElements can be used in in_() clauses now, such as bind
4724
parameters, etc. #476
4725
- reverse operators implemented for `CompareMixin` elements,
4726
allows expressions like "5 + somecolumn" etc. #474
4727
- the "where" criterion of an update() and delete() now correlates
4728
embedded select() statements against the table being updated or
4729
deleted. this works the same as nested select() statement
4730
correlation, and can be disabled via the correlate=False flag on
4731
the embedded select().
4732
- column labels are now generated in the compilation phase, which
4733
means their lengths are dialect-dependent. So on oracle a label
4734
that gets truncated to 30 chars will go out to 63 characters
4735
on postgres. Also, the true labelname is always attached as the
4736
accessor on the parent Selectable so theres no need to be aware
4737
of the "truncated" label names [ticket:512].
4738
- column label and bind param "truncation" also generate
4739
deterministic names now, based on their ordering within the
4740
full statement being compiled. this means the same statement
4741
will produce the same string across application restarts and
4742
allowing DB query plan caching to work better.
4743
- the "mini" column labels generated when using subqueries, which
4744
are to work around glitchy SQLite behavior that doesnt understand
4745
"foo.id" as equivalent to "id", are now only generated in the case
4746
that those named columns are selected from (part of [ticket:513])
4747
- the label() method on ColumnElement will properly propagate the
4748
TypeEngine of the base element out to the label, including a label()
4749
created from a scalar=True select() statement.
4750
- MS-SQL better detects when a query is a subquery and knows not to
4751
generate ORDER BY phrases for those [ticket:513]
4752
- fix for fetchmany() "size" argument being positional in most
4754
- sending None as an argument to func.<something> will produce
4756
- query strings in unicode URLs get keys encoded to ascii
4758
- slight tweak to raw execute() change to also support tuples
4759
for positional parameters, not just lists [ticket:523]
4760
- fix to case() construct to propagate the type of the first
4761
WHEN condition as the return type of the case statement
4763
- fixed critical issue when, after options(eagerload()) is used,
4764
the mapper would then always apply query "wrapping" behavior
4765
for all subsequent LIMIT/OFFSET/DISTINCT queries, even if no
4766
eager loading was applied on those subsequent queries.
4767
- added query.with_parent(someinstance) method. searches for
4768
target instance using lazy join criterion from parent instance.
4769
takes optional string "property" to isolate the desired relation.
4770
also adds static Query.query_from_parent(instance, property)
4771
version. [ticket:541]
4772
- improved query.XXX_by(someprop=someinstance) querying to use
4773
similar methodology to with_parent, i.e. using the "lazy" clause
4774
which prevents adding the remote instance's table to the SQL,
4775
thereby making more complex conditions possible [ticket:554]
4776
- added generative versions of aggregates, i.e. sum(), avg(), etc.
4777
to query. used via query.apply_max(), apply_sum(), etc.
4779
- fix to using distinct() or distinct=True in combination with
4781
- corresponding to label/bindparam name generation, eager loaders
4782
generate deterministic names for the aliases they create using
4784
- improved/fixed custom collection classes when giving it "set"/
4785
"sets.Set" classes or subclasses (was still looking for append()
4786
methods on them during lazy loads)
4787
- restored old "column_property()" ORM function (used to be called
4788
"column()") to force any column expression to be added as a property
4789
on a mapper, particularly those that aren't present in the mapped
4790
selectable. this allows "scalar expressions" of any kind to be
4791
added as relations (though they have issues with eager loads).
4792
- fix to many-to-many relationships targeting polymorphic mappers
4794
- making progress with session.merge() as well as combining its
4795
usage with entity_name [ticket:543]
4796
- the usual adjustments to relationships between inheriting mappers,
4797
in this case establishing relation()s to subclass mappers where
4798
the join conditions come from the superclass' table
4800
- informix support added ! courtesy James Zhang, who put a ton
4803
- removed silly behavior where sqlite would reflect UNIQUE indexes
4804
as part of the primary key (?!)
4806
- small fix to allow successive compiles of the same SELECT object
4807
which features LIMIT/OFFSET. oracle dialect needs to modify
4808
the object to have ROW_NUMBER OVER and wasn't performing
4809
the full series of steps on successive compiles.
4811
- support for SSL arguments given as inline within URL query string,
4812
prefixed with "ssl_", courtesy terjeros@gmail.com.
4813
- mysql uses "DESCRIBE [<schemaname>].<tablename>", catching exceptions
4814
if table doesnt exist, in order to determine if a table exists.
4815
this supports unicode table names as well as schema names. tested
4816
with MySQL5 but should work with 4.1 series as well. (#557)
4818
- big fix to AssociationProxy so that multiple AssociationProxy
4819
objects can be associated with a single association collection.
4820
- assign_mapper names methods according to their keys (i.e. __name__)
4823
- pyodbc is now the preferred DB-API for MSSQL, and if no module is
4824
specifically requested, will be loaded first on a module probe.
4826
- The @@SCOPE_IDENTITY is now used instead of @@IDENTITY. This
4827
behavior may be overridden with the engine_connect
4828
"use_scope_identity" keyword parameter, which may also be specified
4835
- bindparam() names are now repeatable! specify two
4836
distinct bindparam()s with the same name in a single statement,
4837
and the key will be shared. proper positional/named args translate
4838
at compile time. for the old behavior of "aliasing" bind parameters
4839
with conflicting names, specify "unique=True" - this option is
4840
still used internally for all the auto-genererated (value-based)
4843
- slightly better support for bind params as column clauses, either
4844
via bindparam() or via literal(), i.e. select([literal('foo')])
4846
- MetaData can bind to an engine either via "url" or "engine" kwargs
4847
to constructor, or by using connect() method. BoundMetaData is
4848
identical to MetaData except engine_or_url param is required.
4849
DynamicMetaData is the same and provides thread-local connections be
4852
- exists() becomes useable as a standalone selectable, not just in a
4853
WHERE clause, i.e. exists([columns], criterion).select()
4855
- correlated subqueries work inside of ORDER BY, GROUP BY
4857
- fixed function execution with explicit connections, i.e.
4858
conn.execute(func.dosomething())
4860
- use_labels flag on select() wont auto-create labels for literal text
4861
column elements, since we can make no assumptions about the text. to
4862
create labels for literal columns, you can say "somecol AS
4863
somelabel", or use literal_column("somecol").label("somelabel")
4865
- quoting wont occur for literal columns when they are "proxied" into
4866
the column collection for their selectable (is_literal flag is
4867
propagated). literal columns are specified via
4868
literal_column("somestring").
4870
- added "fold_equivalents" boolean argument to Join.select(), which
4871
removes 'duplicate' columns from the resulting column clause that
4872
are known to be equivalent based on the join condition. this is of
4873
great usage when constructing subqueries of joins which Postgres
4874
complains about if duplicate column names are present.
4876
- fixed use_alter flag on ForeignKeyConstraint [ticket:503]
4878
- fixed usage of 2.4-only "reversed" in topological.py [ticket:506]
4880
- for hackers, refactored the "visitor" system of ClauseElement and
4881
SchemaItem so that the traversal of items is controlled by the
4882
ClauseVisitor itself, using the method visitor.traverse(item).
4883
accept_visitor() methods can still be called directly but will not
4884
do any traversal of child items. ClauseElement/SchemaItem now have a
4885
configurable get_children() method to return the collection of child
4886
elements for each parent object. This allows the full traversal of
4887
items to be clear and unambiguous (as well as loggable), with an
4888
easy method of limiting a traversal (just pass flags which are
4889
picked up by appropriate get_children() methods). [ticket:501]
4891
- the "else_" parameter to the case statement now properly works when
4895
- the full featureset of the SelectResults extension has been merged
4896
into a new set of methods available off of Query. These methods
4897
all provide "generative" behavior, whereby the Query is copied
4898
and a new one returned with additional criterion added.
4899
The new methods include:
4901
filter() - applies select criterion to the query
4902
filter_by() - applies "by"-style criterion to the query
4903
avg() - return the avg() function on the given column
4904
join() - join to a property (or across a list of properties)
4905
outerjoin() - like join() but uses LEFT OUTER JOIN
4906
limit()/offset() - apply LIMIT/OFFSET
4907
range-based access which applies limit/offset:
4908
session.query(Foo)[3:5]
4909
distinct() - apply DISTINCT
4910
list() - evaluate the criterion and return results
4912
no incompatible changes have been made to Query's API and no methods
4913
have been deprecated. Existing methods like select(), select_by(),
4914
get(), get_by() all execute the query at once and return results
4915
like they always did. join_to()/join_via() are still there although
4916
the generative join()/outerjoin() methods are easier to use.
4918
- the return value for multiple mappers used with instances() now
4919
returns a cartesian product of the requested list of mappers,
4920
represented as a list of tuples. this corresponds to the documented
4921
behavior. So that instances match up properly, the "uniquing" is
4922
disabled when this feature is used.
4924
- Query has add_entity() and add_column() generative methods. these
4925
will add the given mapper/class or ColumnElement to the query at
4926
compile time, and apply them to the instances() method. the user is
4927
responsible for constructing reasonable join conditions (otherwise
4928
you can get full cartesian products). result set is the list of
4929
tuples, non-uniqued.
4931
- strings and columns can also be sent to the *args of instances()
4932
where those exact result columns will be part of the result tuples.
4934
- a full select() construct can be passed to query.select() (which
4935
worked anyway), but also query.selectfirst(), query.selectone()
4936
which will be used as is (i.e. no query is compiled). works
4937
similarly to sending the results to instances().
4939
- eager loading will not "aliasize" "order by" clauses that were
4940
placed in the select statement by something other than the eager
4941
loader itself, to fix possibility of dupe columns as illustrated in
4942
[ticket:495]. however, this means you have to be more careful with
4943
the columns placed in the "order by" of Query.select(), that you
4944
have explicitly named them in your criterion (i.e. you cant rely on
4945
the eager loader adding them in for you)
4947
- added a handy multi-use "identity_key()" method to Session, allowing
4948
the generation of identity keys for primary key values, instances,
4949
and rows, courtesy Daniel Miller
4951
- many-to-many table will be properly handled even for operations that
4952
occur on the "backref" side of the operation [ticket:249]
4954
- added "refresh-expire" cascade [ticket:492]. allows refresh() and
4955
expire() calls to propagate along relationships.
4957
- more fixes to polymorphic relations, involving proper lazy-clause
4958
generation on many-to-one relationships to polymorphic mappers
4959
[ticket:493]. also fixes to detection of "direction", more specific
4960
targeting of columns that belong to the polymorphic union vs. those
4963
- some fixes to relationship calcs when using "viewonly=True" to pull
4964
in other tables into the join condition which arent parent of the
4965
relationship's parent/child mappings
4967
- flush fixes on cyclical-referential relationships that contain
4968
references to other instances outside of the cyclical chain, when
4969
some of the objects in the cycle are not actually part of the flush
4971
- put an aggressive check for "flushing object A with a collection of
4972
B's, but you put a C in the collection" error condition - **even if
4973
C is a subclass of B**, unless B's mapper loads polymorphically.
4974
Otherwise, the collection will later load a "B" which should be a
4975
"C" (since its not polymorphic) which breaks in bi-directional
4976
relationships (i.e. C has its A, but A's backref will lazyload it as
4977
a different instance of type "B") [ticket:500] This check is going
4978
to bite some of you who do this without issues, so the error message
4979
will also document a flag "enable_typechecks=False" to disable this
4980
checking. But be aware that bi-directional relationships in
4981
particular become fragile without this check.
4984
- options() method on SelectResults now implemented "generatively"
4985
like the rest of the SelectResults methods [ticket:472]. But
4986
you're going to just use Query now anyway.
4988
- query() method is added by assignmapper. this helps with
4989
navigating to all the new generative methods on Query.
4992
- removed seconds input on DATE column types (probably
4993
should remove the time altogether)
4995
- null values in float fields no longer raise errors
4997
- LIMIT with OFFSET now raises an error (MS-SQL has no OFFSET support)
4999
- added an facility to use the MSSQL type VARCHAR(max) instead of TEXT
5000
for large unsized string fields. Use the new "text_as_varchar" to
5001
turn it on. [ticket:509]
5003
- ORDER BY clauses without a LIMIT are now stripped in subqueries, as
5004
MS-SQL forbids this usage
5006
- cleanup of module importing code; specifiable DB-API module; more
5007
explicit ordering of module preferences. [ticket:480]
5010
- got binary working for any size input ! cx_oracle works fine,
5011
it was my fault as BINARY was being passed and not BLOB for
5012
setinputsizes (also unit tests werent even setting input sizes).
5014
- also fixed CLOB read/write on a separate changeset.
5016
- auto_setinputsizes defaults to True for Oracle, fixed cases where
5017
it improperly propagated bad types.
5020
- added a catchall **kwargs to MSString, to help reflection of
5021
obscure types (like "varchar() binary" in MS 4.0)
5023
- added explicit MSTimeStamp type which takes effect when using
5029
- the value of "case_sensitive" defaults to True now, regardless of the
5030
casing of the identifier, unless specifically set to False. this is
5031
because the object might be label'ed as something else which does
5032
contain mixed case, and propigating "case_sensitive=False" breaks that.
5033
Other fixes to quoting when using labels and "fake" column objects
5034
- added a "supports_execution()" method to ClauseElement, so that
5035
individual kinds of clauses can express if they are appropriate for
5036
executing...such as, you can execute a "select", but not a "Table" or a
5038
- fixed argument passing to straight textual execute() on engine,
5039
connection. can handle *args or a list instance for positional, **kwargs
5040
or a dict instance for named args, or a list of list or dicts to invoke
5042
- small fix to BoundMetaData to accept unicode or string URLs
5043
- fixed named PrimaryKeyConstraint generation [ticket:466] courtesy
5045
- fixed generation of CHECK constraints on columns [ticket:464]
5046
- fixes to tometadata() operation to propagate Constraints at column and
5049
- when returning "rowid" as the ORDER BY column or in use with ROW_NUMBER
5050
OVER, oracle dialect checks the selectable its being applied to and will
5051
switch to table PK if not applicable, i.e. for a UNION. checking for
5052
DISTINCT, GROUP BY (other places that rowid is invalid) still a TODO.
5053
allows polymorphic mappings to function, [ticket:436]
5054
- sequences on a non-pk column will properly fire off on INSERT
5055
- added PrefetchingResultProxy support to pre-fetch LOB columns when they
5056
are known to be present, fixes [ticket:435]
5057
- implemented reflection of tables based on synonyms, including across
5058
dblinks [ticket:379]
5059
- issues a log warning when a related table cant be reflected due to
5060
certain permission errors [ticket:363]
5062
- fix to reflection on older DB's that might return array() type for
5063
"show variables like" statements
5065
- better reflection of sequences for alternate-schema Tables [ticket:442]
5066
- sequences on a non-pk column will properly fire off on INSERT
5067
- added PGInterval type [ticket:460], PGInet type [ticket:444]
5069
- preliminary support for pyodbc (Yay!) [ticket:419]
5070
- better support for NVARCHAR types added [ticket:298]
5071
- fix for commit logic on pymssql
5072
- fix for query.get() with schema [ticket:456]
5073
- fix for non-integer relationships [ticket:473]
5074
- DB-API module now selectable at run-time [ticket:419]
5075
- now passes many more unit tests [tickets:422, 481, 415]
5076
- better unittest compatibility with ANSI functions [ticket:479]
5077
- improved support for implicit sequence PK columns with auto-insert
5079
- fix for blank password in adodbapi [ticket:371]
5080
- fixes to get unit tests working with pyodbc [ticket:481]
5081
- fix to auto_identity_insert on db-url query
5082
- added query_timeout to db-url query parms. currently works only for
5084
- tested with pymssql 0.8.0 (which is now LGPL)
5086
- another refactoring to relationship calculation. Allows more accurate
5087
ORM behavior with relationships from/to/between mappers, particularly
5088
polymorphic mappers, also their usage with Query, SelectResults. tickets
5089
include [ticket:439], [ticket:441], [ticket:448].
5090
- removed deprecated method of specifying custom collections on classes;
5091
you must now use the "collection_class" option. the old way was
5092
beginning to produce conflicts when people used assign_mapper(), which
5093
now patches an "options" method, in conjunction with a relationship
5094
named "options". (relationships take precedence over monkeypatched
5095
assign_mapper methods).
5096
- extension() query option propagates to Mapper._instance() method so that
5097
all loading-related methods get called [ticket:454]
5098
- eager relation to an inheriting mapper wont fail if no rows returned for
5100
- eager relation loading bug fixed for eager relation on multiple
5101
descendant classes [ticket:486]
5102
- fix for very large topological sorts, courtesy ants.aasma at gmail
5104
- eager loading is slightly more strict about detecting "self-referential"
5105
relationships, specifically between polymorphic mappers. this results in
5106
an "eager degrade" to lazy loading.
5107
- improved support for complex queries embedded into "where" criterion for
5108
query.select() [ticket:449]
5109
- mapper options like eagerload(), lazyload(), deferred(), will work for
5110
"synonym()" relationships [ticket:485]
5111
- fixed bug where cascade operations incorrectly included deleted
5112
collection items in the cascade [ticket:445]
5113
- fixed relationship deletion error when one-to-many child item is moved
5114
to a new parent in a single unit of work [ticket:478]
5115
- fixed relationship deletion error where parent/child with a single
5116
column as PK/FK on the child would raise a "blank out the primary key"
5117
error, if manually deleted or "delete" cascade without "delete-orphan"
5119
- fix to deferred so that load operation doesnt mistakenly occur when only
5120
PK col attributes are set
5122
- implemented foreign_keys argument to mapper [ticket:385]. use in
5123
conjunction with primaryjoin/secondaryjoin arguments to specify/override
5124
foreign keys defined on the Table instance.
5125
- contains_eager('foo') automatically implies eagerload('foo')
5126
- added "alias" argument to contains_eager(). use it to specify the string
5127
name or Alias instance of an alias used in the query for the eagerly
5128
loaded child items. easier to use than "decorator"
5129
- added "contains_alias()" option for result set mapping to an alias of
5131
- added support for py2.5 "with" statement with SessionTransaction
5134
- added distinct() method to SelectResults. generally should only make a
5135
difference when using count().
5136
- added options() method to SelectResults, equivalent to query.options()
5138
- added optional __table_opts__ dictionary to ActiveMapper, will send kw
5139
options to Table objects [ticket:462]
5140
- added selectfirst(), selectfirst_by() to assign_mapper [ticket:467]
5144
- global "insure"->"ensure" change. in US english "insure" is actually
5145
largely interchangeable with "ensure" (so says the dictionary), so I'm not
5146
completely illiterate, but its definitely sub-optimal to "ensure" which is
5149
- added "fetchmany()" support to ResultProxy
5150
- added support for column "key" attribute to be useable in
5151
row[<key>]/row.<key>
5152
- changed "BooleanExpression" to subclass from "BinaryExpression", so that
5153
boolean expressions can also follow column-clause behaviors (i.e. label(),
5155
- trailing underscores are trimmed from func.<xxx> calls, such as func.if_()
5156
- fix to correlation of subqueries when the column list of the select
5157
statement is constructed with individual calls to append_column(); this
5158
fixes an ORM bug whereby nested select statements were not getting
5159
correlated with the main select generated by the Query object.
5160
- another fix to subquery correlation so that a subquery which has only one
5161
FROM element will *not* correlate that single element, since at least one
5162
FROM element is required in a query.
5163
- default "timezone" setting is now False. this corresponds to Python's
5164
datetime behavior as well as Postgres' timestamp/time types (which is the
5165
only timezone-sensitive dialect at the moment) [ticket:414]
5166
- the "op()" function is now treated as an "operation", rather than a
5167
"comparison". the difference is, an operation produces a BinaryExpression
5168
from which further operations can occur whereas comparison produces the
5169
more restrictive BooleanExpression
5170
- trying to redefine a reflected primary key column as non-primary key raises
5172
- type system slightly modified to support TypeDecorators that can be
5173
overridden by the dialect (ok, thats not very clear, it allows the mssql
5174
tweak below to be possible)
5176
- added an NVarchar type (produces NVARCHAR), also MSUnicode which provides
5177
Unicode-translation for the NVarchar regardless of dialect convert_unicode
5180
- fix to the initial checkfirst for tables to take current schema into
5181
account [ticket:424]
5182
- postgres has an optional "server_side_cursors=True" flag which will utilize
5183
server side cursors. these are appropriate for fetching only partial
5184
results and are necessary for working with very large unbounded result
5185
sets. While we'd like this to be the default behavior, different
5186
environments seem to have different results and the causes have not been
5187
isolated so we are leaving the feature off by default for now. Uses an
5188
apparently undocumented psycopg2 behavior recently discovered on the
5189
psycopg mailing list.
5190
- added "BIGSERIAL" support for postgres table with
5191
PGBigInteger/autoincrement
5192
- fixes to postgres reflection to better handle when schema names are
5193
present; thanks to jason (at) ncsmags.com [ticket:402]
5195
- mysql is inconsistent with what kinds of quotes it uses in foreign keys
5196
during a SHOW CREATE TABLE, reflection updated to accomodate for all three
5198
- mysql table create options work on a generic passthru now, i.e. Table(...,
5199
mysql_engine='InnoDB', mysql_collate="latin1_german2_ci",
5200
mysql_auto_increment="5", mysql_<somearg>...), helps [ticket:418]
5202
- order of constraint creation puts primary key first before all other
5203
constraints; required for firebird, not a bad idea for others [ticket:408]
5204
- Firebird fix to autoload multifield foreign keys [ticket:409]
5205
- Firebird NUMERIC type properly handles a type without precision
5208
- *slight* support for binary, but still need to figure out how to insert
5209
reasonably large values (over 4K). requires auto_setinputsizes=True sent to
5210
create_engine(), rows must be fully fetched individually, etc.
5212
- poked the first hole in the can of worms: saying
5213
query.select_by(somerelationname=someinstance) will create the join of the
5214
primary key columns represented by "somerelationname"'s mapper to the
5215
actual primary key in "someinstance".
5216
- reworked how relations interact with "polymorphic" mappers, i.e. mappers
5217
that have a select_table as well as polymorphic flags. better determination
5218
of proper join conditions, interaction with user- defined join conditions,
5219
and support for self-referential polymorphic mappers.
5220
- related to polymorphic mapping relations, some deeper error checking when
5221
compiling relations, to detect an ambiguous "primaryjoin" in the case that
5222
both sides of the relationship have foreign key references in the primary
5223
join condition. also tightened down conditions used to locate "relation
5224
direction", associating the "foreignkey" of the relationship with the
5226
- a little bit of improvement to the concept of a "concrete" inheritance
5227
mapping, though that concept is not well fleshed out yet (added test case
5228
to support concrete mappers on top of a polymorphic base).
5229
- fix to "proxy=True" behavior on synonym()
5230
- fixed bug where delete-orphan basically didn't work with many-to-many
5231
relationships [ticket:427], backref presence generally hid the symptom
5232
- added a mutex to the mapper compilation step. ive been reluctant to add any
5233
kind of threading anything to SA but this is one spot that its its really
5234
needed since mappers are typically "global", and while their state does not
5235
change during normal operation, the initial compilation step does modify
5236
internal state significantly, and this step usually occurs not at
5237
module-level initialization time (unless you call compile()) but at
5239
- basic idea of "session.merge()" actually implemented. needs more testing.
5240
- added "compile_mappers()" function as a shortcut to compiling all mappers
5241
- fix to MapperExtension create_instance so that entity_name properly
5242
associated with new instance
5243
- speed enhancements to ORM object instantiation, eager loading of rows
5244
- invalid options sent to 'cascade' string will raise an exception
5246
- fixed bug in mapper refresh/expire whereby eager loaders didnt properly
5247
re-populate item lists [ticket:407]
5248
- fix to post_update to ensure rows are updated even for non insert/delete
5249
scenarios [ticket:413]
5250
- added an error message if you actually try to modify primary key values on
5251
an entity and then flush it [ticket:412]
5253
- added "validate=False" argument to assign_mapper, if True will ensure that
5254
only mapped attributes are named [ticket:426]
5255
- assign_mapper gets "options", "instances" functions added (i.e.
5256
MyClass.instances())
5259
- string-based FROM clauses fixed, i.e. select(..., from_obj=["sometext"])
5260
- fixes to passive_deletes flag, lazy=None (noload) flag
5261
- added example/docs for dealing with large collections
5262
- added object_session() method to sqlalchemy namespace
5263
- fixed QueuePool bug whereby its better able to reconnect to a database
5264
that was not reachable (thanks to Sébastien Lelong), also fixed dispose()
5266
- patch that makes MySQL rowcount work correctly! [ticket:396]
5267
- fix to MySQL catch of 2006/2014 errors to properly re-raise OperationalError
5271
- major connection pool bug fixed. fixes MySQL out of sync
5272
errors, will also prevent transactions getting rolled back
5273
accidentally in all DBs [ticket:387]
5274
- major speed enhancements vs. 0.3.1, to bring speed
5275
back to 0.2.8 levels
5276
- made conditional dozens of debug log calls that were
5277
time-intensive to generate log messages
5278
- fixed bug in cascade rules whereby the entire object graph
5279
could be unnecessarily cascaded on the save/update cascade
5280
- various speedups in attributes module
5281
- identity map in Session is by default *no longer weak referencing*.
5282
to have it be weak referencing, use create_session(weak_identity_map=True)
5284
- MySQL detects errors 2006 (server has gone away) and 2014
5285
(commands out of sync) and invalidates the connection on which it occured.
5286
- MySQL bool type fix: [ticket:307]
5287
- postgres reflection fixes: [ticket:349] [ticket:382]
5288
- added keywords for EXCEPT, INTERSECT, EXCEPT ALL, INTERSECT ALL
5290
- assign_mapper in assignmapper extension returns the created mapper
5292
- added label() function to Select class, when scalar=True is used
5293
to create a scalar subquery
5294
i.e. "select x, y, (select max(foo) from table) AS foomax from table"
5295
- added onupdate and ondelete keyword arguments to ForeignKey; propagate
5296
to underlying ForeignKeyConstraint if present. (dont propagate in the
5297
other direction, however)
5298
- fix to session.update() to preserve "dirty" status of incoming object
5299
- sending a selectable to an IN via the in_() function no longer creates
5300
a "union" out of multiple selects; only one selectable to a the in_() function
5301
is allowed now (make a union yourself if union is needed)
5302
- improved support for disabling save-update cascade via cascade="none" etc.
5303
- added "remote_side" argument to relation(), used only with self-referential
5304
mappers to force the direction of the parent/child relationship. replaces
5305
the usage of the "foreignkey" parameter for "switching" the direction.
5306
"foreignkey" argument is deprecated for all uses and will eventually
5307
be replaced by an argument dedicated to ForeignKey specification on mappers.
5311
- some new Pool utility classes, updated docs
5312
- "use_threadlocal" on Pool defaults to False (same as create_engine)
5313
- fixed direct execution of Compiled objects
5314
- create_engine() reworked to be strict about incoming **kwargs. all keyword
5315
arguments must be consumed by one of the dialect, connection pool, and engine
5316
constructors, else a TypeError is thrown which describes the full set of
5317
invalid kwargs in relation to the selected dialect/pool/engine configuration.
5319
- MySQL catches exception on "describe" and reports as NoSuchTableError
5320
- further fixes to sqlite booleans, weren't working as defaults
5321
- fix to postgres sequence quoting when using schemas
5323
- the "delete" cascade will load in all child objects, if they were not
5324
loaded already. this can be turned off (i.e. the old behavior) by setting
5325
passive_deletes=True on a relation().
5326
- adjustments to reworked eager query generation to not fail on circular
5327
eager-loaded relationships (like backrefs)
5328
- fixed bug where eagerload() (nor lazyload()) option didn't properly
5329
instruct the Query whether or not to use "nesting" when producing a
5331
- fixed bug in circular dependency sorting at flush time; if object A
5332
contained a cyclical many-to-one relationship to object B, and object B
5333
was just attached to object A, *but* object B itself wasnt changed,
5334
the many-to-one synchronize of B's primary key attribute to A's foreign key
5335
attribute wouldnt occur. [ticket:360]
5336
- implemented from_obj argument for query.count, improves count function
5337
on selectresults [ticket:325]
5338
- added an assertion within the "cascade" step of ORM relationships to check
5339
that the class of object attached to a parent object is appropriate
5340
(i.e. if A.items stores B objects, raise an error if a C is appended to A.items)
5341
- new extension sqlalchemy.ext.associationproxy, provides transparent
5342
"association object" mappings. new example
5343
examples/association/proxied_association.py illustrates.
5344
- improvement to single table inheritance to load full hierarchies beneath
5346
- fix to subtle condition in topological sort where a node could appear twice,
5348
- additional rework to topological sort, refactoring, for [ticket:365]
5349
- "delete-orphan" for a certain type can be set on more than one parent class;
5350
the instance is an "orphan" only if its not attached to *any* of those parents
5354
- logging is now implemented via standard python "logging" module.
5355
"echo" keyword parameters are still functional but set/unset
5356
log levels for their respective classes/instances. all logging
5357
can be controlled directly through the Python API by setting
5358
INFO and DEBUG levels for loggers in the "sqlalchemy" namespace.
5359
class-level logging is under "sqlalchemy.<module>.<classname>",
5360
instance-level logging under "sqlalchemy.<module>.<classname>.0x..<00-FF>".
5361
Test suite includes "--log-info" and "--log-debug" arguments
5362
which work independently of --verbose/--quiet. Logging added
5363
to orm to allow tracking of mapper configurations, row iteration.
5364
- the documentation-generation system has been overhauled to be
5365
much simpler in design and more integrated with Markdown
5366
- Specific Databases:
5368
- sqlite boolean datatype converts False/True to 0/1 by default
5369
- fixes to Date/Time (SLDate/SLTime) types; works as good as postgres
5372
- fixes bug 261 (table reflection broken for MS-SQL case-sensitive
5374
- can now specify port for pymssql
5375
- introduces new "auto_identity_insert" option for auto-switching
5376
between "SET IDENTITY_INSERT" mode when values specified for IDENTITY columns
5377
- now supports multi-column foreign keys
5378
- fix to reflecting date/datetime columns
5379
- NCHAR and NVARCHAR type support added
5381
- Oracle has experimental support for cx_Oracle.TIMESTAMP, which requires
5382
a setinputsizes() call on the cursor that is now enabled via the
5383
'auto_setinputsizes' flag to the oracle dialect.
5385
- aliases do not use "AS"
5386
- correctly raises NoSuchTableError when reflecting non-existent table
5388
- a fair amount of cleanup to the schema package, removal of ambiguous
5389
methods, methods that are no longer needed. slightly more constrained
5390
useage, greater emphasis on explicitness
5391
- the "primary_key" attribute of Table and other selectables becomes
5392
a setlike ColumnCollection object; is ordered but not numerically
5393
indexed. a comparison clause between two pks that are derived from the
5394
same underlying tables (i.e. such as two Alias objects) can be generated
5395
via table1.primary_key==table2.primary_key
5396
- ForeignKey(Constraint) supports "use_alter=True", to create/drop a foreign key
5397
via ALTER. this allows circular foreign key relationships to be set up.
5398
- append_item() methods removed from Table and Column; preferably
5399
construct Table/Column/related objects inline, but if needed use
5400
append_column(), append_foreign_key(), append_constraint(), etc.
5401
- table.create() no longer returns the Table object, instead has no
5402
return value. the usual case is that tables are created via metadata,
5403
which is preferable since it will handle table dependencies.
5404
- added UniqueConstraint (goes at Table level), CheckConstraint
5405
(goes at Table or Column level).
5406
- index=False/unique=True on Column now creates a UniqueConstraint,
5407
index=True/unique=False creates a plain Index,
5408
index=True/unique=True on Column creates a unique Index. 'index'
5409
and 'unique' keyword arguments to column are now boolean only; for
5410
explcit names and groupings of indexes or unique constraints, use the
5411
UniqueConstraint/Index constructs explicitly.
5412
- added autoincrement=True to Column; will disable schema generation
5413
of SERIAL/AUTO_INCREMENT/identity seq for postgres/mysql/mssql if
5414
explicitly set to False
5415
- TypeEngine objects now have methods to deal with copying and comparing
5416
values of their specific type. Currently used by the ORM, see below.
5417
- fixed condition that occurred during reflection when a primary key
5418
column was explciitly overridden, where the PrimaryKeyConstraint would
5419
get both the reflected and the programmatic column doubled up
5420
- the "foreign_key" attribute on Column and ColumnElement in general
5421
is deprecated, in favor of the "foreign_keys" list/set-based attribute,
5422
which takes into account multiple foreign keys on one column.
5423
"foreign_key" will return the first element in the "foreign_keys" list/set
5424
or None if the list is empty.
5425
- Connections/Pooling/Execution:
5426
- connection pool tracks open cursors and automatically closes them
5427
if connection is returned to pool with cursors still opened. Can be
5428
affected by options which cause it to raise an error instead, or to
5429
do nothing. fixes issues with MySQL, others
5430
- fixed bug where Connection wouldnt lose its Transaction
5431
after commit/rollback
5432
- added scalar() method to ComposedSQLEngine, ResultProxy
5433
- ResultProxy will close() the underlying cursor when the ResultProxy
5434
itself is closed. this will auto-close cursors for ResultProxy objects
5435
that have had all their rows fetched (or had scalar() called).
5436
- ResultProxy.fetchall() internally uses DBAPI fetchall() for better efficiency,
5437
added to mapper iteration as well (courtesy Michael Twomey)
5439
- changed "for_update" parameter to accept False/True/"nowait"
5440
and "read", the latter two of which are interpreted only by
5441
Oracle and Mysql [ticket:292]
5442
- added extract() function to sql dialect
5443
(SELECT extract(field FROM expr))
5444
- BooleanExpression includes new "negate" argument to specify
5445
the appropriate negation operator if one is available.
5446
- calling a negation on an "IN" or "IS" clause will result in
5447
"NOT IN", "IS NOT" (as opposed to NOT (x IN y)).
5448
- Function objects know what to do in a FROM clause now. their
5449
behavior should be the same, except now you can also do things like
5450
select(['*'], from_obj=[func.my_function()]) to get multiple
5451
columns from the result, or even use sql.column() constructs to name the
5452
return columns [ticket:172]
5454
- attribute tracking modified to be more intelligent about detecting
5455
changes, particularly with mutable types. TypeEngine objects now
5456
take a greater role in defining how to compare two scalar instances,
5457
including the addition of a MutableType mixin which is implemented by
5458
PickleType. unit-of-work now tracks the "dirty" list as an expression
5459
of all persistent objects where the attribute manager detects changes.
5460
The basic issue thats fixed is detecting changes on PickleType
5461
objects, but also generalizes type handling and "modified" object
5462
checking to be more complete and extensible.
5463
- a wide refactoring to "attribute loader" and "options" architectures.
5464
ColumnProperty and PropertyLoader define their loading behaivor via switchable
5465
"strategies", and MapperOptions no longer use mapper/property copying
5466
in order to function; they are instead propagated via QueryContext
5467
and SelectionContext objects at query/instances time.
5468
All of the internal copying of mappers and properties that was used to handle
5469
inheritance as well as options() has been removed; the structure
5470
of mappers and properties is much simpler than before and is clearly laid out
5471
in the new 'interfaces' module.
5472
- related to the mapper/property overhaul, internal refactoring to
5473
mapper instances() method to use a SelectionContext object to track
5474
state during the operation.
5475
SLIGHT API BREAKAGE: the append_result() and populate_instances()
5476
methods on MapperExtension have a slightly different method signature
5477
now as a result of the change; hoping that these methods are not
5478
in widespread use as of yet.
5479
- instances() method moved to Query now, backwards-compatible
5480
version remains on Mapper.
5481
- added contains_eager() MapperOption, used in conjunction with
5482
instances() to specify properties that should be eagerly loaded
5483
from the result set, using their plain column names by default, or translated
5484
given an custom row-translation function.
5485
- more rearrangements of unit-of-work commit scheme to better allow
5486
dependencies within circular flushes to work properly...updated
5487
task traversal/logging implementation
5488
- polymorphic mappers (i.e. using inheritance) now produces INSERT
5489
statements in order of tables across all inherited classes
5491
- added an automatic "row switch" feature to mapping, which will
5492
detect a pending instance/deleted instance pair with the same
5493
identity key and convert the INSERT/DELETE to a single UPDATE
5494
- "association" mappings simplified to take advantage of
5495
automatic "row switch" feature
5496
- "custom list classes" is now implemented via the "collection_class"
5497
keyword argument to relation(). the old way still works but is
5498
deprecated [ticket:212]
5499
- added "viewonly" flag to relation(), allows construction of
5500
relations that have no effect on the flush() process.
5501
- added "lockmode" argument to base Query select/get functions,
5502
including "with_lockmode" function to get a Query copy that has
5503
a default locking mode. Will translate "read"/"update"
5504
arguments into a for_update argument on the select side.
5506
- implemented "version check" logic in Query/Mapper, used
5507
when version_id_col is in effect and query.with_lockmode()
5508
is used to get() an instance thats already loaded
5509
- post_update behavior improved; does a better job at not
5510
updating too many rows, updates only required columns
5512
- adjustments to eager loading so that its "eager chain" is
5513
kept separate from the normal mapper setup, thereby
5514
preventing conflicts with lazy loader operation, fixes
5516
- fix to deferred group loading
5517
- session.flush() wont close a connection it opened [ticket:346]
5518
- added "batch=True" flag to mapper; if False, save_obj
5519
will fully save one object at a time including calls
5520
to before_XXXX and after_XXXX
5521
- added "column_prefix=None" argument to mapper; prepends the
5522
given string (typically '_') to column-based attributes automatically
5523
set up from the mapper's Table
5524
- specifying joins in the from_obj argument of query.select() will
5525
replace the main table of the query, if the table is somewhere within
5526
the given from_obj. this makes it possible to produce custom joins and
5527
outerjoins in queries without the main table getting added twice.
5529
- eagerloading is adjusted to more thoughtfully attach its LEFT OUTER JOINs
5530
to the given query, looking for custom "FROM" clauses that may have
5531
already been set up.
5532
- added join_to and outerjoin_to transformative methods to SelectResults,
5533
to build up join/outerjoin conditions based on property names. also
5534
added select_from to explicitly set from_obj parameter.
5535
- removed "is_primary" flag from mapper.
5538
- cleanup on connection methods + documentation. custom DBAPI
5539
arguments specified in query string, 'connect_args' argument
5540
to 'create_engine', or custom creation function via 'creator'
5541
function to 'create_engine'.
5542
- added "recycle" argument to Pool, is "pool_recycle" on create_engine,
5543
defaults to 3600 seconds; connections after this age will be closed and
5544
replaced with a new one, to handle db's that automatically close
5545
stale connections [ticket:274]
5546
- changed "invalidate" semantics with pooled connection; will
5547
instruct the underlying connection record to reconnect the next
5548
time its called. "invalidate" will also automatically be called
5549
if any error is thrown in the underlying call to connection.cursor().
5550
this will hopefully allow the connection pool to reconnect to a
5551
database that had been stopped and started without restarting
5552
the connecting application [ticket:121]
5553
- eesh ! the tutorial doctest was broken for quite some time.
5554
- add_property() method on mapper does a "compile all mappers"
5555
step in case the given property references a non-compiled mapper
5556
(as it did in the case of the tutorial !)
5557
- [ticket:277] check for pg sequence already existing before create
5558
- if a contextual session is established via MapperExtension.get_session
5559
(as it is using the sessioncontext plugin, etc), a lazy load operation
5560
will use that session by default if the parent object is not
5561
persistent with a session already.
5562
- lazy loads will not fire off for an object that does not have a
5563
database identity (why?
5564
see http://www.sqlalchemy.org/trac/wiki/WhyDontForeignKeysLoadData)
5565
- unit-of-work does a better check for "orphaned" objects that are
5566
part of a "delete-orphan" cascade, for certain conditions where the
5567
parent isnt available to cascade from.
5568
- mappers can tell if one of their objects is an "orphan" based
5569
on interactions with the attribute package. this check is based
5570
on a status flag maintained for each relationship
5571
when objects are attached and detached from each other.
5572
- it is now invalid to declare a self-referential relationship with
5573
"delete-orphan" (as the abovementioned check would make them impossible
5575
- improved the check for objects being part of a session when the
5576
unit of work seeks to flush() them as part of a relationship..
5577
- [ticket:280] statement execution supports using the same BindParam
5578
object more than once in an expression; simplified handling of positional
5579
parameters. nice job by Bill Noon figuring out the basic idea.
5580
- postgres reflection moved to use pg_schema tables, can be overridden
5581
with use_information_schema=True argument to create_engine
5582
[ticket:60], [ticket:71]
5583
- added case_sensitive argument to MetaData, Table, Column, determines
5584
itself automatically based on if a parent schemaitem has a non-None
5585
setting for the flag, or if not, then whether the identifier name is all lower
5586
case or not. when set to True, quoting is applied to identifiers with mixed or
5587
uppercase identifiers. quoting is also applied automatically in all cases to
5588
identifiers that are known to be reserved words or contain other non-standard
5589
characters. various database dialects can override all of this behavior, but
5590
currently they are all using the default behavior. tested with postgres, mysql,
5591
sqlite, oracle. needs more testing with firebird, ms-sql. part of the ongoing
5592
work with [ticket:155]
5593
- unit tests updated to run without any pysqlite installed; pool
5594
test uses a mock DBAPI
5595
- urls support escaped characters in passwords [ticket:281]
5596
- added limit/offset to UNION queries (though not yet in oracle)
5597
- added "timezone=True" flag to DateTime and Time types. postgres
5598
so far will convert this to "TIME[STAMP] (WITH|WITHOUT) TIME ZONE",
5599
so that control over timezone presence is more controllable (psycopg2
5600
returns datetimes with tzinfo's if available, which can create confusion
5601
against datetimes that dont).
5602
- fix to using query.count() with distinct, **kwargs with SelectResults
5603
count() [ticket:287]
5604
- deregister Table from MetaData when autoload fails; [ticket:289]
5605
- import of py2.5s sqlite3 [ticket:293]
5606
- unicode fix for startswith()/endswith() [ticket:296]
5609
- quoting facilities set up so that database-specific quoting can be
5610
turned on for individual table, schema, and column identifiers when
5611
used in all queries/creates/drops. Enabled via "quote=True" in
5612
Table or Column, as well as "quote_schema=True" in Table. Thanks to
5613
Aaron Spike for his excellent efforts.
5614
- assignmapper was setting is_primary=True, causing all sorts of mayhem
5615
by not raising an error when redundant mappers were set up, fixed
5616
- added allow_null_pks option to Mapper, allows rows where some
5617
primary key columns are null (i.e. when mapping to outer joins etc)
5618
- modifcation to unitofwork to not maintain ordering within the
5619
"new" list or within the UOWTask "objects" list; instead, new objects
5620
are tagged with an ordering identifier as they are registered as new
5621
with the session, and the INSERT statements are then sorted within the
5622
mapper save_obj. the INSERT ordering has basically been pushed all
5623
the way to the end of the flush cycle. that way the various sorts and
5624
organizations occuring within UOWTask (particularly the circular task
5625
sort) dont have to worry about maintaining order (which they werent anyway)
5626
- fixed reflection of foreign keys to autoload the referenced table
5627
if it was not loaded already
5628
- [ticket:256] - pass URL query string arguments to connect() function
5629
- [ticket:257] - oracle boolean type
5630
- custom primary/secondary join conditions in a relation *will* be propagated
5631
to backrefs by default. specifying a backref() will override this behavior.
5632
- better check for ambiguous join conditions in sql.Join; propagates to a
5633
better error message in PropertyLoader (i.e. relation()/backref()) for when
5634
the join condition can't be reasonably determined.
5635
- sqlite creates ForeignKeyConstraint objects properly upon table
5637
- adjustments to pool stemming from changes made for [ticket:224].
5638
overflow counter should only be decremented if the connection actually
5639
succeeded. added a test script to attempt testing this.
5640
- fixed mysql reflection of default values to be PassiveDefault
5641
- added reflected 'tinyint', 'mediumint' type to MS-SQL [ticket:263],
5643
- SingletonThreadPool has a size and does a cleanup pass, so that
5644
only a given number of thread-local connections stay around (needed
5645
for sqlite applications that dispose of threads en masse)
5646
- fixed small pickle bug(s) with lazy loaders [ticket:265] [ticket:267]
5647
- fixed possible error in mysql reflection where certain versions
5648
return an array instead of string for SHOW CREATE TABLE call
5649
- fix to lazy loads when mapping to joins [changeset:1770]
5650
- all create()/drop() calls have a keyword argument of "connectable".
5651
"engine" is deprecated.
5652
- fixed ms-sql connect() to work with adodbapi
5653
- added "nowait" flag to Select()
5654
- inheritance check uses issubclass() instead of direct __mro__ check
5655
to make sure class A inherits from B, allowing mapper inheritance to more
5656
flexibly correspond to class inheritance [ticket:271]
5657
- SelectResults will use a subselect, when calling an aggregate (i.e.
5658
max, min, etc.) on a SelectResults that has an ORDER BY clause
5660
- fixes to types so that database-specific types more easily used;
5661
fixes to mysql text types to work with this methodology
5663
- some fixes to sqlite date type organization
5664
- added MSTinyInteger to MS-SQL [ticket:263]
5667
- big overhaul to schema to allow truly composite primary and foreign
5668
key constraints, via new ForeignKeyConstraint and PrimaryKeyConstraint
5670
Existing methods of primary/foreign key creation have not been changed
5671
but use these new objects behind the scenes. table creation
5672
and reflection is now more table oriented rather than column oriented.
5674
- overhaul to MapperExtension calling scheme, wasnt working very well
5676
- tweaks to ActiveMapper, supports self-referential relationships
5677
- slight rearrangement to objectstore (in activemapper/threadlocal)
5678
so that the SessionContext is referenced by '.context' instead
5679
of subclassed directly.
5680
- activemapper will use threadlocal's objectstore if the mod is
5681
activated when activemapper is imported
5682
- small fix to URL regexp to allow filenames with '@' in them
5683
- fixes to Session expunge/update/etc...needs more cleanup.
5684
- select_table mappers *still* werent always compiling
5685
- fixed up Boolean datatype
5686
- added count()/count_by() to list of methods proxied by assignmapper;
5687
this also adds them to activemapper
5688
- connection exceptions wrapped in DBAPIError
5689
- ActiveMapper now supports autoloading column definitions from the
5690
database if you supply a __autoload__ = True attribute in your
5691
mapping inner-class. Currently this does not support reflecting
5693
- deferred column load could screw up the connection status in
5694
a flush() under some circumstances, this was fixed
5695
- expunge() was not working with cascade, fixed.
5696
- potential endless loop in cascading operations fixed.
5697
- added "synonym()" function, applied to properties to have a
5698
propname the same as another, for the purposes of overriding props
5699
and allowing the original propname to be accessible in select_by().
5700
- fix to typing in clause construction which specifically helps
5701
type issues with polymorphic_union (CAST/ColumnClause propagates
5702
its type to proxy columns)
5703
- mapper compilation work ongoing, someday it'll work....moved
5704
around the initialization of MapperProperty objects to be after
5705
all mappers are created to better handle circular compilations.
5706
do_init() method is called on all properties now which are more
5707
aware of their "inherited" status if so.
5708
- eager loads explicitly disallowed on self-referential relationships, or
5709
relationships to an inheriting mapper (which is also self-referential)
5710
- reduced bind param size in query._get to appease the picky oracle
5712
- added 'checkfirst' argument to table.create()/table.drop(), as
5713
well as table.exists() [ticket:234]
5714
- some other ongoing fixes to inheritance [ticket:245]
5715
- attribute/backref/orphan/history-tracking tweaks as usual...
5718
- fixed endless loop bug in select_by(), if the traversal hit
5719
two mappers that referenced each other
5720
- upgraded all unittests to insert './lib/' into sys.path,
5721
working around new setuptools PYTHONPATH-killing behavior
5722
- further fixes with attributes/dependencies/etc....
5723
- improved error handling for when DynamicMetaData is not connected
5724
- MS-SQL support largely working (tested with pymssql)
5725
- ordering of UPDATE and DELETE statements within groups is now
5726
in order of primary key values, for more deterministic ordering
5727
- after_insert/delete/update mapper extensions now called per object,
5728
not per-object-per-table
5729
- further fixes/refactorings to mapper compilation
5732
- try/except when the mapper sets init.__name__ on a mapped class,
5734
- fixed bug where threadlocal engine would still autocommit
5735
despite a transaction in progress
5736
- lazy load and deferred load operations require the parent object
5737
to be in a Session to do the operation; whereas before the operation
5738
would just return a blank list or None, it now raises an exception.
5739
- Session.update() is slightly more lenient if the session to which
5740
the given object was formerly attached to was garbage collected;
5741
otherwise still requires you explicitly remove the instance from
5742
the previous Session.
5743
- fixes to mapper compilation, checking for more error conditions
5744
- small fix to eager loading combined with ordering/limit/offset
5745
- utterly remarkable: added a single space between 'CREATE TABLE'
5746
and '(<the rest of it>' since *thats how MySQL indicates a non-
5747
reserved word tablename.....* [ticket:206]
5748
- more fixes to inheritance, related to many-to-many relations
5750
- fixed bug when specifying explicit module to mysql dialect
5751
- when QueuePool times out it raises a TimeoutError instead of
5752
erroneously making another connection
5753
- Queue.Queue usage in pool has been replaced with a locally
5754
modified version (works in py2.3/2.4!) that uses a threading.RLock
5755
for a mutex. this is to fix a reported case where a ConnectionFairy's
5756
__del__() method got called within the Queue's get() method, which
5757
then returns its connection to the Queue via the the put() method,
5758
causing a reentrant hang unless threading.RLock is used.
5759
- postgres will not place SERIAL keyword on a primary key column
5760
if it has a foreign key constraint
5761
- cursor() method on ConnectionFairy allows db-specific extension
5762
arguments to be propagated [ticket:221]
5763
- lazy load bind params properly propagate column type [ticket:225]
5764
- new MySQL types: MSEnum, MSTinyText, MSMediumText, MSLongText, etc.
5765
more support for MS-specific length/precision params in numeric types
5766
patch courtesy Mike Bernson
5767
- some fixes to connection pool invalidate() [ticket:224]
5770
- overhaul to mapper compilation to be deferred. this allows mappers
5771
to be constructed in any order, and their relationships to each
5772
other are compiled when the mappers are first used.
5773
- fixed a pretty big speed bottleneck in cascading behavior particularly
5774
when backrefs were in use
5775
- the attribute instrumentation module has been completely rewritten; its
5776
now a large degree simpler and clearer, slightly faster. the "history"
5777
of an attribute is no longer micromanaged with each change and is
5778
instead part of a "CommittedState" object created when the
5779
instance is first loaded. HistoryArraySet is gone, the behavior of
5780
list attributes is now more open ended (i.e. theyre not sets anymore).
5781
- py2.4 "set" construct used internally, falls back to sets.Set when
5782
"set" not available/ordering is needed.
5783
- fix to transaction control, so that repeated rollback() calls
5784
dont fail (was failing pretty badly when flush() would raise
5785
an exception in a larger try/except transaction block)
5786
- "foreignkey" argument to relation() can also be a list. fixed
5787
auto-foreignkey detection [ticket:151]
5788
- fixed bug where tables with schema names werent getting indexed in
5789
the MetaData object properly
5790
- fixed bug where Column with redefined "key" property wasnt getting
5791
type conversion happening in the ResultProxy [ticket:207]
5792
- fixed 'port' attribute of URL to be an integer if present
5793
- fixed old bug where if a many-to-many table mapped as "secondary"
5794
had extra columns, delete operations didnt work
5795
- bugfixes for mapping against UNION queries
5796
- fixed incorrect exception class thrown when no DB driver present
5797
- added NonExistentTable exception thrown when reflecting a table
5798
that doesnt exist [ticket:138]
5799
- small fix to ActiveMapper regarding one-to-one backrefs, other
5801
- overridden constructor in mapped classes gets __name__ and
5802
__doc__ from the original class
5803
- fixed small bug in selectresult.py regarding mapper extension
5805
- small tweak to cascade_mappers, not very strongly supported
5806
function at the moment
5807
- some fixes to between(), column.between() to propagate typing
5808
information better [ticket:202]
5809
- if an object fails to be constructed, is not added to the
5810
session [ticket:203]
5811
- CAST function has been made into its own clause object with
5812
its own compilation function in ansicompiler; allows MySQL
5813
to silently ignore most CAST calls since MySQL
5814
seems to only support the standard CAST syntax with Date types.
5815
MySQL-compatible CAST support for strings, ints, etc. a TODO
5818
- big improvements to polymorphic inheritance behavior, enabling it
5819
to work with adjacency list table structures [ticket:190]
5820
- major fixes and refactorings to inheritance relationships overall,
5822
- fixed "echo_pool" flag on create_engine()
5823
- fix to docs, removed incorrect info that close() is unsafe to use
5824
with threadlocal strategy (its totally safe !)
5825
- create_engine() can take URLs as string or unicode [ticket:188]
5826
- firebird support partially completed;
5827
thanks to James Ralston and Brad Clements for their efforts.
5828
- Oracle url translation was broken, fixed, will feed host/port/sid
5829
into cx_oracle makedsn() if 'database' field is present, else uses
5830
straight TNS name from the 'host' field
5831
- fix to using unicode criterion for query.get()/query.load()
5832
- count() function on selectables now uses table primary key or
5833
first column instead of "1" for criterion, also uses label "rowcount"
5835
- got rudimental "mapping to multiple tables" functionality cleaned up,
5836
more correctly documented
5837
- restored global_connect() function, attaches to a DynamicMetaData
5838
instance called "default_metadata". leaving MetaData arg to Table
5839
out will use the default metadata.
5840
- fixes to session cascade behavior, entity_name propigation
5841
- reorganized unittests into subdirectories
5842
- more fixes to threadlocal connection nesting patterns
5845
- "pool" argument to create_engine() properly propagates
5846
- fixes to URL, raises exception if not parsed, does not pass blank
5847
fields along to the DB connect string (a string such as
5848
user:host@/db was breaking on postgres)
5849
- small fixes to Mapper when it inserts and tries to get
5850
new primary key values back
5851
- rewrote half of TLEngine, the ComposedSQLEngine used with
5852
'strategy="threadlocal"'. it now properly implements engine.begin()/
5853
engine.commit(), which nest fully with connection.begin()/trans.commit().
5854
added about six unittests.
5855
- major "duh" in pool.Pool, forgot to put back the WeakValueDictionary.
5856
unittest which was supposed to check for this was also silently missing
5857
it. fixed unittest to ensure that ConnectionFairy properly falls out
5859
- placeholder dispose() method added to SingletonThreadPool, doesnt
5861
- rollback() is automatically called when an exception is raised,
5862
but only if theres no transaction in process (i.e. works more like
5864
- fixed exception raise in sqlite if no sqlite module present
5865
- added extra example detail for association object doc
5866
- Connection adds checks for already being closed
5869
- overhaul to Engine system so that what was formerly the SQLEngine
5870
is now a ComposedSQLEngine which consists of a variety of components,
5871
including a Dialect, ConnectionProvider, etc. This impacted all the
5872
db modules as well as Session and Mapper.
5873
- create_engine now takes only RFC-1738-style strings:
5874
driver://user:password@host:port/database
5875
- total rewrite of connection-scoping methodology, Connection objects
5876
can now execute clause elements directly, added explicit "close" as
5877
well as support throughout Engine/ORM to handle closing properly,
5878
no longer relying upon __del__ internally to return connections
5879
to the pool [ticket:152].
5880
- overhaul to Session interface and scoping. uses hibernate-style
5881
methods, including query(class), save(), save_or_update(), etc.
5882
no threadlocal scope is installed by default. Provides a binding
5883
interface to specific Engines and/or Connections so that underlying
5884
Schema objects do not need to be bound to an Engine. Added a basic
5885
SessionTransaction object that can simplistically aggregate transactions
5886
across multiple engines.
5887
- overhaul to mapper's dependency and "cascade" behavior; dependency logic
5888
factored out of properties.py into a separate module "dependency.py".
5889
"cascade" behavior is now explicitly controllable, proper implementation
5890
of "delete", "delete-orphan", etc. dependency system can now determine at
5891
flush time if a child object has a parent or not so that it makes better
5892
decisions on how that child should be updated in the DB with regards to deletes.
5893
- overhaul to Schema to build upon MetaData object instead of an Engine.
5894
Entire SQL/Schema system can be used with no Engines whatsoever, executed
5895
solely by an explicit Connection object. the "bound" methodlogy exists via the
5896
BoundMetaData for schema objects. ProxyEngine is generally not needed
5897
anymore and is replaced by DynamicMetaData.
5898
- true polymorphic behavior implemented, fixes [ticket:167]
5899
- "oid" system has been totally moved into compile-time behavior;
5900
if they are used in an order_by where they are not available, the order_by
5901
doesnt get compiled, fixes [ticket:147]
5902
- overhaul to packaging; "mapping" is now "orm", "objectstore" is now
5903
"session", the old "objectstore" namespace gets loaded in via the
5904
"threadlocal" mod if used
5905
- mods now called in via "import <modname>". extensions favored over
5906
mods as mods are globally-monkeypatching
5907
- fix to add_property so that it propagates properties to inheriting
5908
mappers [ticket:154]
5909
- backrefs create themselves against primary mapper of its originating
5910
property, priamry/secondary join arguments can be specified to override.
5911
helps their usage with polymorphic mappers
5912
- "table exists" function has been implemented [ticket:31]
5913
- "create_all/drop_all" added to MetaData object [ticket:98]
5914
- improvements and fixes to topological sort algorithm, as well as more
5916
- tutorial page added to docs which also can be run with a custom doctest
5917
runner to ensure its properly working. docs generally overhauled to
5918
deal with new code patterns
5919
- many more fixes, refactorings.
5920
- migration guide is available on the Wiki at
5921
http://www.sqlalchemy.org/trac/wiki/02Migration
5924
- some fixes to topological sort algorithm
5925
- added DISTINCT ON support to Postgres (just supply distinct=[col1,col2..])
5926
- added __mod__ (% operator) to sql expressions
5927
- "order_by" mapper property inherited from inheriting mapper
5928
- fix to column type used when mapper UPDATES/DELETEs
5929
- with convert_unicode=True, reflection was failing, has been fixed
5930
- types types types! still werent working....have to use TypeDecorator again :(
5931
- mysql binary type converts array output to buffer, fixes PickleType
5932
- fixed the attributes.py memory leak once and for all
5933
- unittests are qualified based on the databases that support each one
5934
- fixed bug where column defaults would clobber VALUES clause of insert objects
5935
- fixed bug where table def w/ schema name would force engine connection
5936
- fix for parenthesis to work correctly with subqueries in INSERT/UPDATE
5937
- HistoryArraySet gets extend() method
5938
- fixed lazyload support for other comparison operators besides =
5939
- lazyload fix where two comparisons in the join condition point to the
5941
- added "construct_new" flag to mapper, will use __new__ to create instances
5942
instead of __init__ (standard in 0.2)
5943
- added selectresults.py to SVN, missed it last time
5944
- tweak to allow a many-to-many relationship from a table to itself via
5945
an association table
5946
- small fix to "translate_row" function used by polymorphic example
5947
- create_engine uses cgi.parse_qsl to read query string (out the window in 0.2)
5948
- tweaks to CAST operator
5949
- fixed function names LOCAL_TIME/LOCAL_TIMESTAMP -> LOCALTIME/LOCALTIMESTAMP
5950
- fixed order of ORDER BY/HAVING in compile
5953
- support for MS-SQL added courtesy Rick Morrison, Runar Petursson
5954
- the latest SQLSoup from J. Ellis
5955
- ActiveMapper has preliminary support for inheritance (Jeff Watkins)
5956
- added a "mods" system which allows pluggable modules that modify/augment
5957
core functionality, using the function "install_mods(*modnames)".
5958
- added the first "mod", SelectResults, which modifies mapper selects to
5959
return generators that turn ranges into LIMIT/OFFSET queries
5961
- factored out querying capabilities of Mapper into a separate Query object
5962
which is Session-centric. this improves the performance of mapper.using(session)
5963
and makes other things possible.
5964
- objectstore/Session refactored, the official way to save objects is now
5965
via the flush() method. The begin/commit functionality of Session is factored
5966
into LegacySession which is still established as the default behavior, until
5968
- types system is bound to an engine at query compile time, not schema
5969
construction time. this simplifies the types system as well as the ProxyEngine.
5970
- added 'version_id' keyword argument to mapper. this keyword should reference a
5971
Column object with type Integer, preferably non-nullable, which will be used on
5972
the mapped table to track version numbers. this number is incremented on each
5973
save operation and is specifed in the UPDATE/DELETE conditions so that it
5974
factors into the returned row count, which results in a ConcurrencyError if the
5975
value received is not the expected count.
5976
- added 'entity_name' keyword argument to mapper. a mapper is now associated
5977
with a class via the class object as well as an optional entity_name parameter,
5978
which is a string defaulting to None. any number of primary mappers can be
5979
created for a class, qualified by the entity name. instances of those classes
5980
will issue all of their load and save operations through their
5981
entity_name-qualified mapper, and maintain separate a identity in the identity
5982
map for an otherwise equilvalent object.
5983
- overhaul to the attributes system. code has been clarified, and also fixed to
5984
support proper polymorphic behavior on object attributes.
5985
- added "for_update" flag to Select objects
5986
- some fixes for backrefs
5987
- fix for postgres1 DateTime type
5988
- documentation pages mostly switched over to Markdown syntax
5991
- added SQLSession concept to SQLEngine. this object keeps track of retrieving a
5992
connection from the connection pool as well as an in-progress transaction.
5993
methods push_session() and pop_session() added to SQLEngine which push/pop a new
5994
SQLSession onto the engine, allowing operation upon a second connection "nested"
5995
within the previous one, allowing nested transactions. Other tricks are sure to
5996
come later regarding SQLSession.
5997
- added nest_on argument to objectstore.Session. This is a single SQLEngine or
5998
list of engines for which push_session()/pop_session() will be called each time
5999
this Session becomes the active session (via objectstore.push_session() or
6000
equivalent). This allows a unit of work Session to take advantage of the nested
6001
transaction feature without explicitly calling push_session/pop_session on the
6003
- factored apart objectstore/unitofwork to separate "Session scoping" from
6004
"uow commit heavy lifting"
6005
- added populate_instance() method to MapperExtension. allows an extension to
6006
modify the population of object attributes. this method can call the
6007
populate_instance() method on another mapper to proxy the attribute population
6008
from one mapper to another; some row translation logic is also built in to help
6010
- fixed Oracle8-compatibility "use_ansi" flag which converts JOINs to
6011
comparisons with the = and (+) operators, passes basic unittests
6012
- tweaks to Oracle LIMIT/OFFSET support
6013
- Oracle reflection uses ALL_** views instead of USER_** to get larger
6014
list of stuff to reflect from
6015
- fixes to Oracle foreign key reflection [ticket:105]
6016
- objectstore.commit(obj1, obj2,...) adds an extra step to seek out private
6017
relations on properties and delete child objects, even though its not a global
6019
- lots and lots of fixes to mappers which use inheritance, strengthened the
6020
concept of relations on a mapper being made towards the "local" table for that
6021
mapper, not the tables it inherits. allows more complex compositional patterns
6022
to work with lazy/eager loading.
6023
- added support for mappers to inherit from others based on the same table,
6024
just specify the same table as that of both parent/child mapper.
6025
- some minor speed improvements to the attributes system with regards to
6026
instantiating and populating new objects.
6027
- fixed MySQL binary unit test
6028
- INSERTs can receive clause elements as VALUES arguments, not just literal
6030
- support for calling multi-tokened functions, i.e. schema.mypkg.func()
6031
- added J. Ellis' SQLSoup module to extensions package
6032
- added "polymorphic" examples illustrating methods to load multiple object types
6033
from one mapper, the second of which uses the new populate_instance() method.
6034
small improvements to mapper, UNION construct to help the examples along
6035
- improvements/fixes to session.refresh()/session.expire() (which may have
6036
been called "invalidate" earlier..)
6037
- added session.expunge() which totally removes an object from the current
6039
- added *args, **kwargs pass-thru to engine.transaction(func) allowing easier
6040
creation of transactionalizing decorator functions
6041
- added iterator interface to ResultProxy: "for row in result:..."
6042
- added assertion to tx = session.begin(); tx.rollback(); tx.begin(), i.e. cant
6043
use it after a rollback()
6044
- added date conversion on bind parameter fix to SQLite enabling dates to
6046
- improvements to subqueries to more intelligently construct their FROM
6047
clauses [ticket:116]
6048
- added PickleType to types.
6049
- fixed two bugs with column labels with regards to bind parameters: bind param
6050
keynames they are now generated from a column "label" in all relevant cases to
6051
take advantage of excess-name-length rules, and checks for a peculiar collision
6052
against a column named the same as "tablename_colname" added
6053
- major overhaul to unit of work documentation, other documentation sections.
6054
- fixed attributes bug where if an object is committed, its lazy-loaded list got
6055
blown away if it hadnt been loaded
6056
- added unique_connection() method to engine, connection pool to return a
6057
connection that is not part of the thread-local context or any current
6059
- added invalidate() function to pooled connection. will remove the connection
6060
from the pool. still need work for engines to auto-reconnect to a stale DB
6062
- added distinct() function to column elements so you can do
6063
func.count(mycol.distinct())
6064
- added "always_refresh" flag to Mapper, creates a mapper that will always
6065
refresh the attributes of objects it gets/selects from the DB, overwriting any
6069
- create_engine() now uses genericized parameters; host/hostname,
6070
db/dbname/database, password/passwd, etc. for all engine connections. makes
6071
engine URIs much more "universal"
6072
- added support for SELECT statements embedded into a column clause, using the
6074
- another overhaul to EagerLoading when used in conjunction with mappers that
6075
inherit; improvements to eager loads figuring out their aliased queries
6076
correctly, also relations set up against a mapper with inherited mappers will
6077
create joins against the table that is specific to the mapper itself (i.e. and
6078
not any tables that are inherited/are further down the inheritance chain),
6079
this can be overridden by using custom primary/secondary joins.
6080
- added J.Ellis patch to mapper.py so that selectone() throws an exception
6081
if query returns more than one object row, selectfirst() to not throw the
6082
exception. also adds selectfirst_by (synonymous with get_by) and selectone_by
6083
- added onupdate parameter to Column, will exec SQL/python upon an update
6084
statement.Also adds "for_update=True" to all DefaultGenerator subclasses
6085
- added support for Oracle table reflection contributed by Andrija Zaric;
6086
still some bugs to work out regarding composite primary keys/dictionary selection
6087
- checked in an initial Firebird module, awaiting testing.
6088
- added sql.ClauseParameters dictionary object as the result for
6089
compiled.get_params(), does late-typeprocessing of bind parameters so
6090
that the original values are easier to access
6091
- more docs for indexes, column defaults, connection pooling, engine construction
6092
- overhaul to the construction of the types system. uses a simpler inheritance
6093
pattern so that any of the generic types can be easily subclassed, with no need
6095
- added "convert_unicode=False" parameter to SQLEngine, will cause all String
6096
types to perform unicode encoding/decoding (makes Strings act like Unicodes)
6097
- added 'encoding="utf8"' parameter to engine. the given encoding will be
6098
used for all encode/decode calls within Unicode types as well as Strings
6099
when convert_unicode=True.
6100
- improved support for mapping against UNIONs, added polymorph.py example
6101
to illustrate multi-class mapping against a UNION
6102
- fix to SQLite LIMIT/OFFSET syntax
6103
- fix to Oracle LIMIT syntax
6104
- added backref() function, allows backreferences to have keyword arguments
6105
that will be passed to the backref.
6106
- Sequences and ColumnDefault objects can do execute()/scalar() standalone
6107
- SQL functions (i.e. func.foo()) can do execute()/scalar() standalone
6108
- fix to SQL functions so that the ANSI-standard functions, i.e. current_timestamp
6109
etc., do not specify parenthesis. all other functions do.
6110
- added settattr_clean and append_clean to SmartProperty, which set
6111
attributes without triggering a "dirty" event or any history. used as:
6112
myclass.prop1.setattr_clean(myobject, 'hi')
6113
- improved support to column defaults when used by mappers; mappers will pull
6114
pre-executed defaults from statement's executed bind parameters
6115
(pre-conversion) to populate them into a saved object's attributes; if any
6116
PassiveDefaults have fired off, will instead post-fetch the row from the DB to
6117
populate the object.
6118
- added 'get_session().invalidate(*obj)' method to objectstore, instances will
6119
refresh() themselves upon the next attribute access.
6120
- improvements to SQL func calls including an "engine" keyword argument so
6121
they can be execute()d or scalar()ed standalone, also added func accessor to
6123
- fix to MySQL4 custom table engines, i.e. TYPE instead of ENGINE
6124
- slightly enhanced logging, includes timestamps and a somewhat configurable
6125
formatting system, in lieu of a full-blown logging system
6126
- improvements to the ActiveMapper class from the TG gang, including
6127
many-to-many relationships
6128
- added Double and TinyInt support to mysql
6131
- completed "post_update" feature, will add a second update statement before
6132
inserts and after deletes in order to reconcile a relationship without any
6133
dependencies being created; used when persisting two rows that are dependent
6135
- completed mapper.using(session) function, localized per-object Session
6136
functionality; objects can be declared and manipulated as local to any
6137
user-defined Session
6138
- fix to Oracle "row_number over" clause with multiple tables
6139
- mapper.get() was not selecting multiple-keyed objects if the mapper's table was a join,
6140
such as in an inheritance relationship, this is fixed.
6141
- overhaul to sql/schema packages so that the sql package can run all on its own,
6142
producing selects, inserts, etc. without any engine dependencies. builds upon
6143
new TableClause/ColumnClause lexical objects. Schema's Table/Column objects
6144
are the "physical" subclasses of them. simplifies schema/sql relationship,
6145
extensions (like proxyengine), and speeds overall performance by a large margin.
6146
removes the entire getattr() behavior that plagued 0.1.1.
6147
- refactoring of how the mapper "synchronizes" data between two objects into a
6148
separate module, works better with properties attached to a mapper that has an
6149
additional inheritance relationship to one of the related tables, also the same
6150
methodology used to synchronize parent/child objects now used by mapper to
6151
synchronize between inherited and inheriting mappers.
6152
- made objectstore "check for out-of-identitymap" more aggressive, will perform the
6153
check when object attributes are modified or the object is deleted
6154
- Index object fully implemented, can be constructed standalone, or via
6155
"index" and "unique" arguments on Columns.
6156
- added "convert_unicode" flag to SQLEngine, will treat all String/CHAR types
6157
as Unicode types, with raw-byte/utf-8 translation on the bind parameter and
6159
- postgres maintains a list of ANSI functions that must have no parenthesis so
6160
function calls with no arguments work consistently
6161
- tables can be created with no engine specified. this will default their engine
6162
to a module-scoped "default engine" which is a ProxyEngine. this engine can
6163
be connected via the function "global_connect".
6164
- added "refresh(*obj)" method to objectstore / Session to reload the attributes of
6165
any set of objects from the database unconditionally
6168
- fixed a recursive call in schema that was somehow running 994 times then returning
6169
normally. broke nothing, slowed down everything. thanks to jpellerin for finding this.
6172
- small fix to Function class so that expressions with a func.foo() use the type of the
6173
Function object (i.e. the left side) as the type of the boolean expression, not the
6174
other side which is more of a moving target (changeset 1020).
6175
- creating self-referring mappers with backrefs slightly easier (but still not that easy -
6177
- fixes to one-to-one mappings (changeset 1015)
6178
- psycopg1 date/time issue with None fixed (changeset 1005)
6179
- two issues related to postgres, which doesnt want to give you the "lastrowid"
6180
since oids are deprecated:
6181
* postgres database-side defaults that are on primary key cols *do* execute
6182
explicitly beforehand, even though thats not the idea of a PassiveDefault. this is
6183
because sequences on columns get reflected as PassiveDefaults, but need to be explicitly
6184
executed on a primary key col so we know what we just inserted.
6185
* if you did add a row that has a bunch of database-side defaults on it,
6186
and the PassiveDefault thing was working the old way, i.e. they just execute on
6187
the DB side, the "cant get the row back without an OID" exception that occurred
6188
also will not happen unless someone (usually the ORM) explicitly asks for it.
6189
- fixed a glitch with engine.execute_compiled where it was making a second
6190
ResultProxy that just got thrown away.
6191
- began to implement newer logic in object properities. you can now say
6192
myclass.attr.property, which will give you the PropertyLoader corresponding to that
6193
attribute, i.e. myclass.mapper.props['attr']
6194
- eager loading has been internally overhauled to use aliases at all times. more
6195
complicated chains of eager loads can now be created without any need for explicit
6196
"use aliases"-type instructions. EagerLoader code is also much simpler now.
6197
- a new somewhat experimental flag "use_update" added to relations, indicates that
6198
this relationship should be handled by a second UPDATE statement, either after a
6199
primary INSERT or before a primary DELETE. handles circular row dependencies.
6200
- added exceptions module, all raised exceptions (except for some
6201
KeyError/AttributeError exceptions) descend from these classes.
6202
- fix to date types with MySQL, returned timedelta converted to datetime.time
6203
- two-phase objectstore.commit operations (i.e. begin/commit) now return a
6204
transactional object (SessionTrans), to more clearly indicate transaction boundaries.
6205
- Index object with create/drop support added to schema
6206
- fix to postgres, where it will explicitly pre-execute a PassiveDefault on a table
6207
if it is a primary key column, pursuant to the ongoing "we cant get inserted rows
6208
back from postgres" issue
6209
- change to information_schema query that gets back postgres table defs, now
6210
uses explicit JOIN keyword, since one user had faster performance with 8.1
6211
- fix to engine.process_defaults so it works correctly with a table that has
6212
different column name/column keys (changset 982)
6213
- a column can only be attached to one table - this is now asserted
6214
- postgres time types descend from Time type
6215
- fix to alltests so that it runs types test (now named testtypes)
6216
- fix to Join object so that it correctly exports its foreign keys (cs 973)
6217
- creating relationships against mappers that use inheritance fixed (cs 973)